基于BIOMOD的黄河源区黄帚橐吾分布  

Distribution of Ligularia virgaurea in the source region of the Yellow River based on BIOMOD

作　　者：黄波 宜树华[1] 张欣雨 马青山 向波[3] 杜嘉星 马建海 孙义 HUANG Bo;YI Shuhua;ZHANG Xinyu;MA Qingshan;XIANG Bo;DU Jiaxing;MA Jianhai;SUN Yi(Institute of Fragile Eco-environment,School of Geographic Science,Nantong University,Nantong 226007,Jiangsu,China;Grassland Station of Huangnan Tibetan Autonomous Prefecture of Qinghai Province,Longwu 811300,Qinghai,China;Chongqing Climate Center,Chongqing 401147,China;PLA Dalian Naval Academy,Dalian 116018,Liaoning,China;Enforcement and supervisory bureau of agriculture and animal husbandry of Huangnan Tibetan Autonomous Prefecture of Qinghai Province,Longwu 811300,Qinghai,China)

机构地区：[1]南通大学脆弱生态环境研究所/地理科学学院,江苏南通226007 [2]青海省黄南州草原站,青海隆务81I300 [3]重庆市气候中心,重庆401147 [4]中国人民解放军海军大连舰艇学院,辽宁大连116018 [5]青海省黄南州农牧综合行政执法监督局,青海隆务811300

出　　处：《草业科学》2020年第11期2198-2210,共13页Pratacultural Science

基　　金：青海省自然科学基金资助项目(2019-ZJ-7054);国家重点研发计划项目(2017YFA0604801);国家自然科学基金资助项目(31901393)。

摘　　要：黄河源区位于青藏高原东北部,属生态环境脆弱的高寒区。黄帚橐吾(Ligularia virgaurea)广泛分布于黄河源区,是限制该区域草地畜牧业发展、导致该区域系统稳定性和服务功能下降的最主要毒杂草之一。目前鲜有黄帚橐吾区域尺度时空分布的研究,而缺乏大尺度、高精度、高效率的监测手段是其主要限制因素。本研究利用团队自主开发的基于无人机的协同航拍和分析系统(Fragmentation Monitoring and Analysis with Aerial Photography,FragMAP)对黄河源区黄帚橐吾的分布进行基础调查,获取有效样本(航拍照片)5000余个并提取黄帚橐吾存在/不存在数据。基于生物多样性模拟(biodiversity modelling,BIOMOD)模型集成平台,采用10种不同模型模拟该区域的黄帚橐吾潜在分布格局及其对气候变化的响应。结果表明:1)BIOMOD能降低预测的不确定性和误差,提高预测精度;2)广义增强回归模型(generalised boosting model,GBM)预测黄河源区黄帚橐吾当前潜在分布效果最好,而随机森林(random forest model,RF)模型则能更好地预测其未来分布,决定黄帚橐吾空间分布的主要因素有小生境[归一化植被指数(normalized difference vegetation index,NDVI)最大值、平均值和范围]、年平均辐射、最冷季辐射、最湿季辐射、最湿期降水等;3)气温升高会导致黄帚橐吾分布范围增加,尤其是源区中部偏东南区域。本研究模拟了黄河源区黄帚橐吾的时空分布特征,为进一步开展其致害等级评价、合理利用和适时防控等工作提供数据支撑,也为该区域草地生态系统和畜牧业的可持续发展提供科学依据。同时,本研究利用FragMAP系统获取大量标准统一、时间序列长、尺度大、精度高的航拍照片,可为BIOMOD模型提供充足、有效的数据基础,从而为草地生态系统生物多样性的研究和保护提供了新方法和新手段。The source region of the Yellow River(SRYR)lies in the northeastern Qinghai-Tibetan Plateau.The SRYR is a typical alpine region that is fragile and vulnerable.Ligularia virgaurea,one of the main toxic weeds,is widely distributed in the SRYR.There are few studies on the spatial-temporal distribution of L.virgaurea at a regional scale,and the most important limiting factor is the lack of broad scale,precise,and efficient monitoring methods.We proposed a practical method based on unmanned aerial vehicles(UAVs)and a widely used species distribution model,BIOdiversity MODelling(BIOMOD),and tested it to determine the distribution of L.virgaurea in the SRYR.During the growing season of 2018,we set 208 working points using the self-developed software-fragmentation monitoring and analysis with aerial photography(FragMAP)and obtained more than 5000 aerial photographs in the SRYR.Based on the data,we simulated the distribution of L.virgaurea using the 10 models in BIOMOD.The results showed that:1)BIOMOD could reduce the uncertainty and improve the prediction performance of L.virgaurea distribution;2)the generalized boosting model exhibited the best performance in predicting the potential distribution of L.virgaurea in SRYR,while the random forest model was the best at predicting its distribution in the future;the main factors affecting its distribution were the microhabitat(maximum value,average value,and ranges of normalized difference vegetation index),annual average radiation,radiation in cold and wet season,and precipitation in the wet season;and 3)climate warming may increase the distribution of L.virgaurea in SRYR,especially in the southeastern central area.This study described the spatial-temporal distribution of L.virgaurea,which provides basic data for estimating the damage rating and for implementing prevention measures in a rational and timely manner.Furthermore,it could provide a scientific basis for the sustainable development of grassland ecosystems and animal husbandry in the SRYR.Meanwhile,the UAVs-based mon

关 键 词：物种分布模型 FragMAP 无人机 气候变化 草地退化 

分 类 号：S812.6[农业科学—草业科学]