Spatial flows of ecosystem services under future climate and land-use changes

Increasing global pressure on natural capital requires sustainable solutions to ensure long-term provision of ecosystem services (ES) which have complex flow dynamics. Although the importance of ES flows has been increasingly recognized during the past years, little is known about how they will be affected by future climate and land-use changes. Here, we integrated ES flows into a scenario-based assessment framework, and evaluated it by clarifying the local and distant effects of ES flows in the Black River basin, China. The spatiotemporal variations of ES flows were investigated by considering different land policies and climate scenarios, and the drivers of ES flow variations were explored. We found increasing inter-regional ES flows toward 2100. Compared to land-use change, the intra-regional flows were more sensitive to climate change, with greater shifts of flow size and synergistic areas identified under a moderate climatic forcing scenario with sustainable management strategies. Precipitation directly affected water retention flow, yet affected flood mitigation flow both directly and indirectly through changing vegetation cover, which was the major driver of soil retention flow. Biodiversity mediated the positive effects of precipitation and vegetation cover on soil retention flow. Our study highlights the importance of embracing the spatiotemporal features of ES flows in sustainable transboundary management and adaptation strategies.


Introduction
Ecosystem services (ES) are realized from ecosystem functions and processes, and bridge nature and human society through flowing from sources to beneficiaries (Diaz et al 2018, IPBES 2019).Since areas with high ES capacity are not usually coincided with the location of human demand, ES variations in one region can have significant effects on the ecosystems, biodiversity and socioeconomic outcomes elsewhere (Schroter et al 2018, Kleemann et al 2020).With increasing globalization, the interactions of goods, information and services via telecoupling processes have been intensified (Liu et al 2016).Therefore, ES flows are associated with transboundary sustainability challenges through changing the distribution of ecosystem benefits and broader considerations of interregional dependances and equity (Koellner et al 2018).
Conserving and managing natural capital require a systematic understanding of the whole provision chain (supply, demand and flow) of ES (Kleemann et al 2020).Quantification of ES flows entails the identification of locations with intensive interactions between ecosystems and human society by explicitly illustrating the flow zones and the difference between theoretical and actual ES provision.The concept of ES flows offers an opportunity to differ theoretical ES capacity from actual ES provision that meet human demand by flow-connected systems (Bagstad et al 2014).Understanding the spatiotemporal variations of ES flows in the face of environmental changes can inform decision-makers about how they can best allocate resources to maintain and enhance human well-being through systematic conservation planning.
Climate change and land-use change are the primary drivers of ES degradation (Kalantari et al 2023).Climate change can affect ES provision by altering the physiological features (e.g.plant growth, metabolism rates) of biomes and the biogeochemical processes (e.g.precipitation, runoff) of ecosystems (Duku andHein 2023, Kalantari et al 2023).Conversion of land-use types with redistributing spatial patterns can change the structure and function of ecosystems by modifying land surface and its supportive relationships with human society (Hasan et al 2020).The effects of climate and land-use changes are likely to become increasingly important as these two phenomena are expected to intensify with rapid population growth and urbanization around the world (IPCC 2023).However, evaluating the effects of climate and land-use changes is often challenged by their interactions and the uncertainties caused by the variations in their impacts over a long timescale.Despite these challenges, integrating climate and land-use changes into ES flow assessments is vital, because management efforts to ensure sustainable and long-term ES provision which ignore these effects could lead to perverse outcomes (Liu 2023).Previous studies have evaluated the effects of climate and landuse changes on ES by focusing on their supply capacity (Langerwisch et al 2018, Pham et al 2019, Sun et al 2023).However, little is known about how ES flows vary with changing climate and land use patterns.
The importance of ES flows has been increasingly recognized during the past years (Wang et al 2022, Liu 2023).Given their political and practical relevance, ES flows have been incorporated into assessments for provisioning (Fridman and Kissinger 2018, Liu et al 2022, Chen et al 2023) and regulating services (Bagstad et al 2019, Kleemann et al 2020, Zhang et al 2023) at regional and national scales.However, these studies mainly focused on the current and/or historical variations of ES flows.There is little knowledge on their future patterns, especially in the context of global changes.To date, the driving mechanism of ES flows remains poorly understood.
To bridge the knowledge gap, we integrated ES flows into a scenario-based assessment framework, and evaluated it by clarifying the local and distant effects of ES flows in the Black River basin, China.The main objectives were to: (1) investigate the spatiotemporal variations of ES flows in the past and under future climate and land-use changes; (2) identify the main drivers by exploring their causal relationships with ES flows; (3) evaluate the variations of the synergies and trade-offs among multiple service flows.Results of this study could provide a basis for transboundary ES payments and systematic conservation planning to ensure sustainable and long-term ES provision.

Study area
The Black River basin is situated in the southern Shaanxi Province, China, with the Qinling Mountains to the south (supplementary figure S1).This region has a temperate continental monsoon climate, with an average annual precipitation of 570 mm which is mainly concentrated between June and September, and an average annual temperature of 13.3 • C. The elevation varies with a tough topography differing from mountainous areas in the south to plains in the north.The basin is home to 0.6 million people, approximately 4.3% of Xi'an's population.However, this region, with an average annual division of 420 million m 3 water resources, is the largest freshwater source for Xi'an city (Wu et al 2023).Past decades have witnessed urbanization with rapid instrument development, as well as intensified floods and soil erosion along the waterways (DWR 2022).These challenges are compounded by the accelerating climate change which affects biotic and abiotic communities in complex ways, threatening the ecological security of the basin.

Projection of climate and land-use changes scenarios
Our analyses contained the projection of different climate and land-use changes scenarios at the near (2030), middle (2050) and end (2100) time nodes of the 21st century (figure 1), and the assessments of ES flows with a combination of empirical and Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) models (Ouyang et al 2016, Sharp et al 2018).We briefly described the scenarios as below, and provided details in supplementary materials.
Future land-use patterns were projected with a Cellular Automat-Markov model, which addresses the suitability of different land uses to biophysical and anthropogenic pressure (Zhang et al 2022).The model ran with a conversion matrix for different land uses and a suitability map which was created by spatially overlaying eight biophysical and socioeconomic factors (supplementary figure S2).The model was validated by comparing the simulated and observed land-use patterns of 2020.On this basis, three land-use patterns under different scenarios were simulated for 2030.(1) Business as usual (BAU).The historical (2010-2020) trend of land-use change was assumed to continue without any policy interventions.(2) Ecological protection priority (EPP).Socioeconomic development was assumed to give priority to ecological protection to maximize ecosystem benefits.Under this scenario, ecological lands (i.e.forest, grassland) were expanded by decreasing their loss rates by 30%.(3) Economic development priority (EDP).This scenario assumed that the adoption of policy measures that accelerate urbanization and economic growth would be prioritized.The conversion rates of ecological lands into impervious surface were increased by 30%.The changing rates of different land uses under the EPP and EDP scenarios were determined according to the Innovation-driven Development Plan for Shaanxi (GOG 2021).Future land-use patterns of 2050 and 2100 were predicted with a baseline year of 2030.
Average annual data of temperature and precipitation during the periods of 2021-2040, 2041-2060 and 2081-2100 were obtained from the shared socioeconomic pathway (SSP) ensembles and used for future projections at the near, middle and end time-horizons, respectively.The scenarios selected as simulations of the potential spectrum of future climate include moderate (SSP245) and extreme (SSP585) forcing scenarios with respectively moderate and rapid socioeconomic growth (Riahi et al 2017).To capture the local variability of future climate conditions, the climatic data were downscaled using a delta method (Navarro-Racines et al 2020) (supplementary figure S3).
Based on the climate and land-use projections, a total of six combined scenarios were obtained for each time node (2030, 2050, 2100) (supplementary figure S4), which allowed us to examine how ES flows would vary under future climate and land-use changes.

Assessments of ES supply, demand and spatial flows
Quantitative assessments were conducted for three services, including water retention, flood mitigation and soil retention, given their data availability and critical roles in ensuring the ecological security of the basin.Additionally, we assessed biodiversity by mapping habitat quality with the InVEST Habitat Quality module which has been adopted in previous studies (Sharp et al 2018, Nematollahi et al 2020).The module determines habitat quality for each pixel by considering its suitability and sensitivity to anthropogenic threats including urbanization, agricultural intensity, road density and industrial activities (supplementary figure S5).Details about the models were given in supplementary tables S1 and S2.
We determined the capacity of water retention by subtracting evapotranspiration and surface runoff from precipitation.The demand capacity of freshwater was calculated as the sum of water consumptions by agricultural, industrial and residential activities, which were spatially visualized with the distribution of cropland, gross domestic product and human population, respectively (Zhang et al 2023).For flood mitigation, the supply capacity was determined as the amounts of storm runoff retained by vegetation, which represent the capacity of ecosystems to intercept, absorb and detain floodwater.The demand area was identified by intersecting the location of 50 years floodplains with developed grid cells (Watson et al 2019).We multiplied the runoff depth for each cell by its area to yield a ranking of demand capacity for flood mitigation service.For soil retention, the supply capacity was determined as the amounts of surface soil retained by vegetation and was calculated as the difference of soil loss under current vegetation condition with those under an extreme scenario where all landscapes were converted into bare land.We determined the amounts of soil that were exported to streams using the InVEST Soil Delivery Ratio module, and identified the demand area as locations with non-zero soil exports.The avoided soil exports by vegetation were assigned as the demand capacity for soil retention service.
Considering the flow process of a service across distances, we defined ES flows as a physical term that connects the ES supply and demand areas (Assis et al 2023).The flow size was calculated as the actual uses of ES from both local and upstream areas (supplementary figure S6).To quantify ES flows, we calculated service surplus as the difference between supply and demand at a pixel level.After satisfying local demand, the surplus with a positive value was assumed to move by following topographical characteristics.We considered pixels with a positive aggregation of supply-demand balance as locations that could protect downstream beneficiaries by exporting ES according to hydrologic connections.We calculated the amounts of ES that were delivered from a source location to its downstream location as the surplus in the source location weighted by the averaged evapotranspiration coefficients for the two locations.The directions in which ES flows move were considered as water flow directions delineated from a digital elevation model.The services for each grid cell were delivered by water flows according to their flow directions until the surplus was negative or reached the export of the basin.To evaluate the scaledependence of ES flows, we quantified inter-regional ES flows that move from the basin to surrounding regions, and intra-regional ES flows that move from upstream pixels to downstream pixels within the basin.The boundaries were considered as the geographical boundary of the basin for the inter-regional flows and a pixel for the intra-regional flows.Future changes of ES flows were evaluated by incorporating the spatial layers of climatic factors and land-use patterns under different scenarios into the assessment models.As a result, 18 maps of future ES flows were obtained for each service (supplementary figure S7).

Statistical analyses
The temporal changes of ES flows were examined with a non-parametric Mann-Kendal test, which detects significantly monotonic direction and measures how the trend varies consistently (Hamed and Ramachandra 1998).The changing rates of ES flows over time were calculated using a Theil-Sen's slope estimator, which has advantages in determining how much the trend increases or decreases in a noisy series (Sen 1968, Theil 1992).To investigate the causal relationship between environmental drivers and ES flows, a structural equation model was constructed using the lavaan package in R 3.5.0(Rosseel 2012).The model was framed by considering ES flows and different drivers relating to climate (precipitation, temperature), topography (elevation, slope), vegetation and soil properties (vegetation cover, biodiversity, soil erodibility) and social environment (population count, agriculture intensity).We fitted the model with a backward stepwise procedure based on an Akaike information criterion (AIC) and the p-value of each regressor term (Fan et al 2016).Iteration was processed by removing the least significant term sequentially.The most parsimonious model with the lowest AIC was retained.The model fitness with evaluation indices of comparative fit index (CFI = 0.913) and root mean square error of approximation (RMSEA = 0.040) suggested an overall good performance (Shipley 2016).

Inter-regional ES flows
Compared to the quantity that met the local demand, larger proportions of services with 52.3% for water retention, 65.5% for flood mitigation and 60.1% for soil retention, flowed out of the basin in 2020.The inter-regional flows increased by 13.5%-37.9%during 2000-2020, with the most increases observed for flood mitigation.The inter-regional flows were predicted to increase toward 2100 (figure 2), with higher amplification observed for flood mitigation, which was estimated to increase by 9.2% (SSP585-EDP) to 16.9% (SSP245-EPP) in relation to 2020.With a same land policy, the SSP245 scenario had overall higher levels of inter-regional flows than the SSP585 scenario.

Intra-regional ES flows
With a baseline of 2020, a consistently spatial heterogeneity was found for the future trends of intra-regional flows, which showed increases in the southern areas, but decreases in the northern areas (figures 3(a) and (b)).Changes of the intra-regional flows in 2100 relative to the baseline year revealed larger differences in the spatial pattern across climate scenarios compared to those across different land-use change scenarios.Under a same climatic forcing, the EPP scenario had only an average growth of <10% in the areas with increased ES flows compared to the BAU and EDP scenarios.However, these areas accounted for over two thirds of the study area under the EPP-SSP245 scenario, and increased by 37.3% (water retention), 69.2% (flood mitigation) and 42.6% (soil retention) compared to those identified under the EPP-SSP585 scenario.Additionally, increased retention of surface water in 2100 was accompanied by increased intraregional flows of flood mitigation and soil retention services, and such synergies peaked under the SSP245-EPP scenario which had the largest increase in the areas overlapped with amplified spatial flows for all services (figure 3(c)).Despite weakening synergies with increased radiative forcing from SSP245 to SSP585 for all service combinations, we determined increases in synergistic areas across the future   scenarios, suggesting that tradeoffs diminished over time and shifted toward synergies.

Effects of climate and land-use changes on ES flows
Precipitation with greater total effects than other factors appeared to be the most important driver of the intra-regional flows for water retention and flood mitigation services, while vegetation cover was identified as the major driver of soil retention flow (figure 4 and supplementary table S3).Positive effects were determined for precipitation on the intra-regional flows water retention and flood mitigation services, and for temperature the soil retention flow.Precipitation affected flood mitigation flow directly and indirectly through vegetation cover.The effects of precipitation on soil retention flow were mediated by vegetation cover and biodiversity.Temperature showed indirect effects on flood mitigation flow through vegetation cover, and affected soil retention flow both directly and indirectly through vegetation cover, biodiversity and soil erodibility.Additionally, the only significantly direct effects were found for population count on water retention flow and for topographical factors, such as elevation on biodiversity and slope on agriculture intensity.

Discussion
Given the rapid urbanization and climate change around the world, how to counteract global pressure on natural capital and reconcile ecological conservation with socioeconomic development have become a central issue in decision-making about sustainable management (Kalantari et al 2023).Understanding the spatiotemporal variations of ES flows over a long timescale and revealing their associations with environmental drivers is fundamental for policy intervention to achieve sustainable and long-term ES provision in the face of global changes.

Interpretation of major findings
To prevent ES degradation, the national government has launched series of ecological restoration projects (e.g.Grain to Green Program, Natural Forest Conservation Program) since the end of last century, which have achieved great increases of vegetation cover and concomitant ecosystem benefits during the past 20 years (Chen et al 2022).The enhancement of ES capacity was attributed to the successful implementation of these restoration projects, which contributed to overall increases of inter-regional flows by altering the biophysical and climatic conditions (Duku and Hein 2023).Increases in the inter-regional flows suggested an intensive telecoupling process and indicated an increasingly important role played by the local ecosystems of the basin in supporting distant ES demand.
While previous studies have investigated the effects of climate and land-use changes by focusing on ES supply capacity (Langerwisch et al 2018, Pham et al 2019, Sun et al 2023), our study further extended this hotspot by providing spatially explicit information on how ES flows would vary with changing climate and land-use patterns.The greater spatial heterogeneity of ES flow variations across different climate scenarios indicated stronger effects of climatic factors than land-use change.This is in line with previous studies which reported high importance of climate-induced changes for future ES provision (Pham et al 2019, Weiskopf et al 2020, Sun et al 2023).The higher increases of spatial flows under the SSP245 scenario than the SSP585 scenario stemmed partially from the varying climatic factors.Although the warming climate could accelerate evapotranspiration and increase the intensity of extreme weather events, the increased precipitation could favor the metabolism and growth of species communities, which offset the negative effects of temperature rises (Duku and Hein 2023).
Future scenarios revealed both positive and negative effects of climate and land-use changes on the intra-regional flows.For a given location, the varying trends with a same or opposite directions for different ES flows indicated an evolvement of the synergy and trade-off relationships among the ES flows.This has a similarity with previous studies where changes in ES interactions have been found over timescales (Qiu et al 2018, Schirpke et al 2019).Although enhancing climate-change adaptivity by improving multiple ES concurrently remains challenging, our study found that the SSP245-EPP scenario had the highest overlap of the areas with increased intra-regional flows for all suggesting a possibility to achieve multiple management goals through minimizing ES trade-offs.

Management implications
Understanding how ES flows will be affected by future climate and land-use changes could help decisionmakers to evaluate past management strategies and counteract undesirable influences (Liu 2023), thus increasing the capacity to develop adaptive strategies to global changes.This knowledge could route from assessments of past land dynamics and future projections by anticipating policy outcomes (Hasan et al 2020, Mandle et al 2021).Considering the time-lag effects of land policies that affect processes and ES provision (Requena-Mullor et al 2018), assessments of ES flows over a long timescale offer an opportunity for decision-makers to capture the slow ecological and social processes which are usually featured by longer timescale than monitoring programs.
The spatial heterogeneity of intra-regional flows, routing partly from the prevailing patterns of biological and geophysical properties (Kalantari et al 2023), underscores the potential of and finescale management practices.Our spatially explicit maps of ES flows could inform decision-makers to identify the areas susceptible to future changes in order to maximize the benefits of policy interventions.While large-scales assessments could mask biophysical and geographical variations which may have differences in the magnitude and/or direction with large-scale changes (Qiu et 2018), understanding ES flows across distinct locations or areas with disproportionate importance could shed lights on how to best allocate resources to improve ES provision at finer scales.Additionally, our study emphasizes the importance of incorporating ES flow interactions and their evolvement over time into management strategies.For example, the implementation of ecological conservation policy under the EPP scenario led to increased flows of water retention, flood mitigation and soil retention services when future climatic forcing is relatively moderate.This illustrates the opportunities of co-managing and enhancing these essential services together through telecoupling processes (Liu 2023).Moreover, the discrepancy of synergistic areas among different land-use scenarios suggested an important role played by land policies in mediating the climatic effects on the interactions among multiple ES flows.This has a similarity with previous studies where the trends of ES supply became more complicated when considering land surface and topographic characteristics (e.g.elevation, vegetation) for climatic effect analyses (Langerwisch et al 2018, Schirpke et al 2019, Weiskopf et al 2020).These results highlight the importance of considering both climate and land-use changes to assess the evolvement of ES interactions along the whole provision chain of ES.

Research limitations and prospects
There are several limitations regarding to the generalization of our results.First, the land projections ignored the slow succession processes of natural ecosystems and their time-lag responses to land policies (Requena-Mullor et al 2018, Schirpke et al 2020).The uncertainties of socioeconomic projections could increase with extending time due to the difficulties in predicting the continuous turnover of policies along the time horizon (Chen et al 2020).Second, ES have diverse dynamic features, depending on service type and spatial scale (Bagstad et al 2014).Extending our study to examine how other types of ES flows response to future environmental changes would be fruitful and more informative to sustainable management.Third, a time-step of years was adopted in this study.However, environmental drivers have seasonal or monthly fluctuations.Detailed spatial data with finer temporal resolutions permit a deeper understanding of ES flows.Moreover, uncertainties could arise from the use of a same variable for assessing ES supply and for checking the drivers of ES flows, which could lead to an overestimation of the effects on ES flows.Although we explored the drivers by focusing on their interactions and indirect effects on ES flows, application of the results should be cautious.More efforts on the driving mechanism of ES flows with systematically dynamic models would favor an in-depth understanding of how ES flows vary under environmental constraints.

Conclusions
This study combined ES flows with scenario analysis to investigate how future climate and land-use changes affect ES flows at pixel and basin scales.The results showed increasing inter-regional ES flows toward 2100, especially under a moderate climatic forcing scenario with sustainable management strategies, which favored the improvements of intraregional flows and the enhancement of ES synergies.Climatic factors affected the intra-regional ES flows far more than land-use change.Our results could benefit local decision-making about conservation investment and the establishment of crossregional compensation mechanisms by identifying where ES flows are sensitive to climate and land-use changes and how the spatial flows would vary over a long time-horizon.The framework of this study could be applied in other regions experiencing similar stresses for sustaining multiple services by feasibly updating datasets and considering the uncertainties caused by system succession, service type, temporal resolution and driving forces.Further understanding of the spatial heterogeneity of the effects of climate and land-use changes on ES flows will be fruitful avenues for future studies.

Figure 1 .
Figure 1.Conceptual diagram of this study.

Figure 2 .
Figure 2. Inter-regional flows of (a) water retention, (b) flood mitigation, and (c) soil retention services in the past (2000-2020) and under different future scenarios.

Figure 3 .
Figure 3. Changes of intra-regional flows within the basin.(a) Historical trends from 2000 to 2020.(b) Spatial changes of future flows in 2100 relative to 2020.(c) Changes in the areas overlapped with increased flows for multiple services.The icons in (c) represent different services illustrated in (b).

Figure 4 .
Figure 4. Relationships between different drivers and ES flows based on the structural equation model.Path are standardized and solid paths indicate significantly direct effects at a significance level of 0.05.Black arrows represent positive effects and red arrows represent negative effects.The width of paths correlates with the magnitude of path coefficients.
ecosystem services in an agricultural landscape Ecol.Appl.28 119-34 Requena-Mullor J M, Quintas-Soriano C, Brandt J, Cabello J and Castro A J 2018 Modeling how land use legacy affects the provision of ecosystem services in Mediterranean southern Spain Environ.Res.Lett.13 114008 Riahi K et al 2017 The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: an overview Glob.Environ.Change 42 153-68 Rosseel Y 2012 lavaan: an R package for structural equation modeling J. Stat.Softw.48 1-36 Schirpke U, Candiago S, Egarter Vigl L, Jäger H, Labadini A, Marsoner T, Meisch C, Tasser E and Tappeiner U 2019 Integrating supply, flow and demand to enhance the understanding of interactions among multiple ecosystem services Sci.Total Environ.651 928-41 Schirpke U, Tscholl S and Tasser E 2020 Spatio-temporal changes in ecosystem service values: effects of land-use changes from past to future (1860-2100) J. Environ.Manage.272 111068 Schroter M et al 2018 Interregional flows of ecosystem services: concepts, typology and four cases Ecosyst.Serv.31 231-41 Sen P K 1968 Estimates of the regression coefficient based on Kendall's Tau J. Am.Stat.Assoc.63 1379-89 Sharp R et al 2018 InVEST user's guide (The Natural Capital Project, Stanford University, University of Minnesota, The Nature Conservancy and World Wildlife Fund) (available at: http://data.naturalcapitalproject.org/invest-releases/3.5.0/ userguide) Shipley B 2016 Cause and Correlation in Biology: A User's Guide to Path Analysis, Structural Equations and Causal Inference (Cambridge University Press) pp 17-32 Sun L, Yu H, Sun M and Wang Y 2023 Coupled impacts of climate and land use changes on regional ecosystem services J. Environ.Manage.326 116753 Theil H 1992 A rank-invariant method of linear and polynomial regression analysis Henri Theil's Contributions to Economics and Econometrics.Advanced Studies in Theoretical and Applied Econometrics vol 23, ed B Raj and J Koerts (Springer) Wang L, Zheng H, Chen Y, Ouyang Z and Hu X 2022 Systematic review of ecosystem services flow measurement: main concepts, methods, applications and future directions Ecosyst.Serv.58 101479 Watson K B, Galford G L, Sonter L J, Koh I and Ricketts T H 2019 Effects of human demand on conservation planning for biodiversity and ecosystem services Conserv.Biol.33 942-52 Weiskopf S R et al 2020 Climate change effects on biodiversity, ecosystems, ecosystem services, and natural resource management in the United States Sci.Total Environ.733 137782 Wu Q, Song J, Sun H, Huang P, Jing K, Xu W, Wang H and Liang D 2023 Spatiotemporal variations of water conservation function based on EOF analysis at multi time scales under different ecosystems of Heihe River Basin J. Environ.Manage.325 116532 Zhang Y, Wang Y, Sun S and Chen X 2023 Quantifying interregional flows of ecosystem services to enhance water security in the Yellow River Basin, China J. Water Resour.Plan.Manage.149 04023018 Zhang Y, Wu T, Song C, Hein L, Shi F, Han M and Ouyang Z 2022 Influences of climate change and land use change on the interactions of ecosystem services in China's Xijiang River Basin Ecosyst.Serv.58 101489