Monitoring the invasion of an exotic tree (Ligustrum lucidum) from 1983 to 2006 with Landsat TM/ETM plus satellite data and Support Vector Machines in Cordoba, Argentina

REMOTE SENSING OF ENVIRONMENT

卷:122

Article

Gavier-Pizarro, Gregorio I.¹  Kuemmerle, Tobias^2,8  Hoyos, Laura E.^3,4  Stewart, Susan I.⁵  Huebner, Cynthia D.⁶  Keuler, Nicholas S.⁷  Radeloff, Volker C.¹ 

[1] Univ Wisconsin, Dept Forest & Wildlife Ecol, Madison, WI 53706 USA

[2] Potsdam Inst Climate Impact Res PIK, D-14412 Potsdam, Germany

[3] Univ Nacl Cordoba, Inst Multidisciplinario Biol Vegetal IMBIV CONICE, RA-5000 Cordoba, Argentina

[4] Univ Nacl Cordoba, Fac Ciencias Exactas Fis & Nat, RA-5000 Cordoba, Argentina

[5] US Forest Serv, USDA, No Res Stn, Evanston, IL 60201 USA

[6] US Forest Serv, USDA, No Res Stn, Morgantown, WV 26505 USA

[7] Univ Wisconsin, Dept Stat, Madison, WI 53706 USA

[8] Humboldt Univ, Dept Geog, D-10099 Berlin, Germany

关键词: Exotic tree invasions;    Non-native invasive species;    Ligustrum lucidum;    Change detection;    Support Vector Machines classification;    Urban expansion;    Landsat Thematic Mapper;    Argentina;    Sierras Chicas;    Cordoba;   

DOI  :  10.1016/j.rse.2011.09.023

来源: Elsevier

【 摘 要 】

In central Argentina, the Chinese tree glossy privet (Ligustrum lucidum) is an aggressive invasive species replacing native forests, forming dense stands, and is thus a major conservation concern. Mapping the spread of biological invasions is a necessary first step toward understanding the factors determining invasion patterns. Urban areas may function as propagule sources for glossy privet because it has been used as a landscaping tree for over a century. The objectives of this paper were to 1) map the patterns of glossy privet expansion from 1983 to 2006 using a time series of Landsat TM/ETM + images, and 2) analyze the spatial pattern of glossy privet stands with regard to urban extent. Using six summer Landsat TM images (1983, 1987, 1992, 1997, 2001, and 2006) the expansion of glossy privet was analyzed using Support Vector Machines (SVM), a non-parametric classifier which we applied to a stack of all images simultaneously, a novel approach in its application to monitor non-native tree invasions. We then measured the area of glossy privet in a series of 200-m buffers at increasing distances around urban areas in 1983 and 2006, and compared it with the amount of privet expected in proportion to buffer area Glossy privet in the study area has spread very rapidly during the 23 years that we studied and the SVM resulted in highly accurate classifications (Kappa Index 0.88, commission error 0.07, omission error 0.16). Between 1983 and 2006 glossy privet area increased 50 times (from 50 to 2500 ha), and 20% of all forest in the study area is now dominated by glossy privet Most of the glossy privet dominated stands were located within 600 m of urban areas. However, the rate of glossy privet expansion accelerated substantially after 1992 and new glossy privet dominated stands tend to be located away from urban areas. This suggests that glossy privet is now self-sustaining, but expected urban growth in the area could further foster glossy privet invasion. Management and development plans should include mitigation efforts to contain this species and prevent invasion into native forests, and citizens should be informed about the risk of invasion associated with the use of glossy privet for landscaping. (C) 2012 Elsevier Inc. All rights reserved.

【 授权许可】

Free