当前鲜有专门针对我国自然保护地的基于可持续发展目标（Sustainable Development Goals, SDGs）的土地利用模拟研究及以此为基础的生态系统服务价值（Ecosystem Service Value, ESV）预测工作。大兴安岭生物多样性优先区是我国自然保护地格局优化的阶段性成果，定量评估区域内的ESV的现状值和未来值，对于规划我国保护优先区域的可持续发展具有重要参考意义。当前面向SDGs的土地利用模拟及ESV研究主要集中于省级行政区划或国家尺度。此外，本研究发现基于合并后的一级地类评估ESV可能会掩盖掉一些关键信息，例如本研究发现针叶林会受到阔叶林的蚕食，这些在前人研究中是鲜有涉及的。

本研究提出一种综合SDGs、系统动力学模型（System Dynamic, SD）和未来土地利用模拟模型（Future Land Use Simulation, FLUS）的SDG-SD-FLUS集成耦合模型，使用该模型获得了2030年五种情景下的ESV。首先，基于2001-2015年的分辨率为300m的欧空局气候变化倡议土地覆盖项目的土地覆被数据（Climate Change Initiative-Land Cover Project of the European Space Agency, ESA CCI-LC），分析土地利用的结构变化、转移规律，提取退化像元分布，在此基础上使用遥感修正后的当量因子法评估历史时期的ESV，并探讨不同生态系统服务功能之间的权衡协同度（Ecosystem Services Trade-off Degree, ESTD）。其次，本研究围绕SDG8.1.1、SDG2.4.1、SDG11.6.2和SDG15.3.1，使用SDG-SD-FLUS对研究区的2020年土地利用进行了模拟，将其与2020年CCI-LC产品进行比较，来判断模型可靠性。最后基于SDG-SD-FLUS模型输出的2030年参考情景、2030年经济发展情景、2030年环境友好情景、2030年粮食生产情景和2030年可持续发展情景的土地利用格局来预测未来时期的ESV。研究结果如下：

（1）本研究构建的SDG-SD-FLUS模型对土地利用数量模拟精度和空间配置精度均较高。将SD模块的2001-2015年输出结果与2001-2015年ESA CCI-LC产品的各地类面积进行比较，得到的相对误差δ总体小于3%，土地利用数量需求模拟精度较好；使用2020年ESA CCI-LC数据作为验证数据，FLUS模块对2020年土地利用格局的模拟结果Kappa系数为0.96，总体精度OA为0.98，土地利用格局空间配置精度较好。

（2）历史时期土地转移规律和基于SDG-SD-FLUS的多情景模拟结果均发现了针叶林的丧失。林地贡献了研究区土地的82.10%，针叶林占林地总构成的92.49%。根据历史时期的土地转移分析发现，2001-2015年针叶林呈现出像元数量的净损失，约23412个像元净转出为其他类型，其中66.39%的针叶林转化为了草原。基于退化像元的识别结果发现，区域面临较高的土地退化风险，总退化面积为196,902公顷。基于SDG-SD-FLUS的模拟结果发现，在所有情景下这种净损失在未来都将持续，未来针叶林将一定程度上转化为阔叶林，表现为在2030年参考情景、2030年经济发展情景、2030年环境友好情景、2030年粮食生产情景、2030年可持续发展情景的转入为新增阔叶林的像元中，针叶林分别贡献了在73.30%，74.36%，70.11%，73.45%和73.92%。本研究发现针叶林在14种地类中具有最高的ESV敏感性系数0.688（Coefficient of Sensitivity, CS），其数量深刻影响ESV总量。

（3）基于现状时期ESV评估结果，以优先区划定年份（2010年）为节点，比较了2001-2010、2010-2015两时段的ESTD变化。经计算，在11种生态系统服务功能之间共形成了110组成对比较关系。在2001-2010年，共有110组ESTD>0，协同率为100%，2010-2015年只有90组ESTD>0，协同率下降到了81.82%。气候调节服务同其他服务功能之间的关系出现了更大的权衡，不利于区域的可持续发展，这可能与发现的针叶林的丧失有关，有研究显示针叶林具有良好的森林抗风性、降雨截留等气候调节服务功能，而研究区的针叶林丧失是一个不可逆转的趋势，这一问题需要引起重视。

（4）在SDG-SD-FLUS模拟的五种情景土地利用的基础上，测算了2020-2030年的ESTD值大小，综合ESV评估结果和土地利用分析规划了研究区未来可行的土地利用方案。研究发现2030年参考情景、2030年经济发展情景、2030年环境友好情景、2030年粮食生产情景、2030年可持续发展情景的ESTD值>0的组合数分别为74组、62组、110组、50组、74组，协同率分别为67.27%、56.36%、100%、45.45%、67.27%。尽管2030年环境友好情景的协同率为100%，但针叶林转阔叶林的数量也是各情景中最多的；可持续发展情景虽然协同率只有67.27%，其数值与2030年参考情景相同，但该情景降低了32组生态系统服务功能对之间的权衡率，限制了土地退化的继续发生，原生植被针叶林在各情景中转出数量相对较少，旱地扩张蚕食其他生态用地的数量最少。在综合考量森林保育、旱地扩张、权衡协同度、ESV评估和敏感性检验等结果的基础上，认为未来该区域应参考2030可持续发展情景进行合理规划。

There are few Sustainable Development Goals (SDGs)-based land use simulation as well as Ecosystem Service Value (ESV) simulation studies specifically aimed at Protected Areas (PAs) in China. Daxing’anling Priority Area for Biodiversity Conservation (PABC) is a stage result of optimizing the PA layout in China. Quantitative assessment of the present and future values of ESV in this study area is of great significance for planning sustainable development of PAs and PABCs. At present, the land use simulation and ESV research of coupled SDGs mainly focus on provincial or national scale. In addition, this study found that the first-level land cover (LC) based ESV assessment may conceal some key information, such as the discovery of coniferous forest encroachment by broad-leaved forest in this study, which was rarely involved in previous studies.

In this study, we proposed SDG-SD-FLUS model that integrated SDGs, System Dynamics (SD) models, and Future Land Use Simulation (FLUS) models, and used the model to obtain ESV under five scenarios in 2030. Firstly, based on the 300 m LC data of the Climate Change Initiative-Land Cover Project of the European Space Agency (ESA CCI-LC) from 2001 to 2015, the structural change and transfer matrix of land use were analyzed, and the degradation pixel distribution was extracted. On this basis, the remote sensing modified equivalent factor method was used to evaluate the ESV in the historical period, and the Ecosystem Services Trade-off Degree (ESTD) between different ecosystem service functions were discussed. Secondly, focusing on SDG 8.1.1, SDG 2.4.1, SDG 11.6.2 and SDG 15.3.1, this study used SDG-SD-FLUS to simulate the land use in 2020 and compare it with the CCI-LC product in 2020 to judge the model reliability. Finally, the ESV in the future period was evaluated based on the land use output of SDG-SD-FLUS in 2030 reference scenario (REF), 2030 economic development scenario (ECO), 2030 environment-friendly scenario (ENV), 2030 food production scenario (FOD) and 2030 sustainable development scenario (SUS). The results of the study are as follows:

(1) The SDG-SD-FLUS model constructed in this study has high precision in both quantitative simulation and spatial layout of land use. In this study, an SDG-SD-FLUS model for SDGs was constructed. Experimental results showed that SDG-SD-FLUS could effectively capture the historical evolution law of land use pattern in the study area. Comparing the output results of the SD module in terms of the pixel area of each land use category from 2001 to 2015 with the pixel area of ESA CCI-LC products from 2001 to 2015, the relative error δ of the SD module was generally less than 3%, indicating that the simulation accuracy of land use quantity demand was good. Using ESA CCI-LC data in 2020 as validation data, the simulation results of FLUS module for land use pattern in 2020 had a Kappa coefficient of 0.96 and an overall accuracy of 0.98, indicating a good spatial simulation accuracy of land use layout in the study area.

(2) The loss of coniferous forests was found in both the historical analysis and the future simulation. Forest land accounted for 82.10% of the land in the study area, and coniferous forest accounted for 92.49% of the total forest land. We found that from 2001 to 2015, the coniferous forest showed a net loss, about 23,412 cells were transferred to other types, of which 66.39% of the coniferous forest was converted into grassland. The study region faced a high risk of land degradation, with a total degraded area of 196,902 hectares. The simulation results based on SDG-SD-FLUS show that the net loss of coniferous forests will continue in the future under all scenarios, and the coniferous forest will be converted into broad-leaved forest to a certain extent in the future. It shows that coniferous forests contribute 73.30%, 74.36%, 70.11%, 73.45% and 73.92% of the new broad-leaved forests in the 2030 REF scenario, 2030 ECO scenario, 2030 ENV scenario, 2030 FOD scenario and 2030 SUS scenario, respectively. In this study, it was found that coniferous forests had the highest ESV Coefficient of Sensitivity (CS) among the 14 land types (CS=0.688), and their number profoundly affected the total amount of ESV.

(3) Based on the results of ESV assessment in the current period, the changes of the ESTD between the two periods of 2001-2010 and 2010-2015 were compared, taking the year of priority area demarcation (2010) as the change point. A total of 110 pairs of 11 ecosystem services were calculated, and the size of the ESTD value changed from 2001 to 2015. From 2001 to 2010, the synergistic rate was 100%, and from 2010 to 2015, the synergistic rate dropped to 81.82%. There is a greater trade-off between climate regulation services and other ecosystem services, which is not conducive to the sustainable development of the region. This may be related to the loss of coniferous forests, which have been shown to have good forest wind resistance, rainfall interception and other climate regulation services, while the loss of coniferous forests in the study area is an irreversible trend, which needs to be paid attention to.

(4) Based on the five scenarios simulated by SDG-SD-FLUS, ESTD from 2020 to 2030 were calculated. This study found that the synergistic rate of the 2030 REF scenario, 2030 ECO scenario, 2030 ENV scenario, 2030 FOD scenario and 2030 SUS scenario were 67.27%, 56.36%, 100%, 45.45% and 67.27%, respectively. Although the synergistic rate of the 2030 ENV scenario in 2030 is 100%, the number of coniferous forests to broad-leaved forests is also the largest among the scenarios. Despite the fact that the synergistic rate of the 2030 SUS scenario is only 67.27%, which is the same value as the 2030 REF scenario, the non-synergistic rate of the 16 groups of ecosystem service function pairs has reduced. The 2030 SUS scenario restricts the continuous occurrence of land degradation. In 2030 SUS scenario, the amount of native vegetation coniferous forest is relatively small, and the amount of dry land expansion encroaches on other ecological land is the least. On the basis of comprehensive consideration of forest conservation, dryland expansion, ESTD calculation result, ESV assessment and CS sensitivity test, it is considered that the region should be rationally planned according to the 2030 SUS scenario for its future.