Fires and land-cover change in the tropics: a remote sensing analysis at the landscape scale

Authors
Citation
H. Eva et Ef. Lambin, Fires and land-cover change in the tropics: a remote sensing analysis at the landscape scale, J BIOGEOGR, 27(3), 2000, pp. 765-776
Citations number
84
Categorie Soggetti
Environment/Ecology
Journal title
JOURNAL OF BIOGEOGRAPHY
ISSN journal
03050270 → ACNP
Volume
27
Issue
3
Year of publication
2000
Pages
765 - 776
Database
ISI
SICI code
0305-0270(200005)27:3<765:FALCIT>2.0.ZU;2-7
Abstract
Aim Fires are an important landscape disturbance which interact in a comple x way with land-use and land-cover change. The objective of this study is t o understand the role of fires in vegetation-cover change and, conversely, the role of land use as a controlling factor of fires. Location The study sites are located in Mate Grosso, Brazil, in Central Afr ica and on the border between Kenya and Tanzania. Methods The role of vegetation fire is addressed through a landscape-scale analysis of the spatial association between maps of active fires and maps o f land-cover change derived from remote sensing data for the different site s in Africa and South America. Results The empirical results of this study clearly support the idea that f ires have widely varying impacts on land cover in savanna and forest ecosys tems. Fires play different roles within the different components of landsca pe mosaics and at different times of the land-cover change trajectory. The impact of fires on vegetation is mainly controlled by land use. Conclusions There is thus a need to consider the socioeconomic purpose of b iomass burning and the context in which such activities are undertaken. In forest ecosystems, a statistically significant relationship exists between the occurrence of fires and forest-cover changes. One could not conclude ho wever, that fires are always the cause of the change in land cover, nor tha t fires are a reliable indicator of 'hot spots' of deforestation. Current l ow spatial resolution information on fire activity derived from remote sens ing systems can be prone to inaccuracies due to a poor co-location of fire with respect to land-cover data, and temporal sampling problems affecting f ire data.