Humanizing Savanna Biogeography: Linking Human Practices with Ecological Patterns in a Frequently Burned Savanna of Southern Mali

Abstract

The savanna biome is defined by the codominance of grasses and trees, the precise ratio of which is a function of numerous variables, both natural and anthropogenic. Fire is a critical determinant of the tree-to-grass ratio because it can suppress tree establishment and growth. Despite the fact that nearly all fires have human causes, most savanna fire ecology models are based solely on ecological variables and fail to integrate human land and fire use patterns. To understand how human burning practices, land cover patterns, and fire regimes are linked I combined the results of an image-based burn scar analysis and land cover maps generated from a Landsat database covering thirty years with interview and survey results. I documented the spatially explicit fire regime for an area of Sudan savanna in southern Mali and linked it to human practices and land cover patterns. The study concludes that burning creates a seasonal mosaic in which some patches regularly burn early and others late, whereas some areas rarely if ever burn. Ecologists attribute mosaic fire regimes to natural factors such as plant succession, but this research demonstrates that the regime is a function of the combined effects of edaphic conditions, vegetation patterns, land use, and widespread burning practices. The anthropogenic burning regime documented here differs from the one based on ecological theory in that its spatiotemporal pattern is relatively consistent from year to year. The findings demonstrate the need to “humanize” savanna models by breaking nature–society dichotomies and synthesizing natural and social science explanations.

La definición del bioma de sabana descansa en el dominio compartido del espacio por hierbas y árboles, cuya razón precisa es una función de numerosas variables, tanto naturales como antropogénicas. El fuego es un determinante crítico de la razón árbol/hierba porque aquél puede suprimir la aparición y crecimiento de árboles. A pesar del hecho de que casi todos los incendios tienen causas humanas, la mayoría de los modelos ecológicos sobre incendios sabaneros se basan solo en variables ecológicas y fallan en integrar los patrones culturales de uso del suelo y utilización del fuego. Para entender cómo se entrelazan las prácticas humanas de quema, los patrones de cubierta del suelo y los regímenes de incendios, en este estudio combiné los resultados del análisis de una cicatriz de fuego registrada en imágenes remotas, màs los mapas de cubierta del suelo generados a partir de una base de datos Landsat que se extiende por treinta años, con los resultados de entrevistas y levantamientos de campo. Documenté el régimen de incendios espacialmente explícito para un área de la sabana sudanesa de la parte sur de Mali, y lo relacioné con prácticas humanas y patrones de cubierta del suelo. El estudio concluye que las quemas crean un mosaico estacional en el que algunos parches con cierta regularidad se queman temprano y otros tardíamente, en tanto que algunas áreas rara vez, o nunca, son quemadas. Los ecólogos atribuyen los regímenes de incendios en mosaico a factores naturales, tales como la sucesión vegetal, pero la presente investigación demuestra que tal régimen es una función de los efectos combinados de condiciones edáficas, patrones de vegetación, uso del suelo y de las bien extendidas prácticas de quema. El régimen antropogénico de quemas documentado en este estudio difiere del que se basa en la teoría ecológica en que su patrón espacio-temporal es relativamente consistente, año tras año. Los descubrimientos demuestran la necesidad de “humanizar” los modelos de sabana, rompiendo con la dicotomía naturaleza-sociedad y sintetizando las explicaciones que provienen de las ciencias naturales y sociales.

Keywords:

• Africa
• fire regime
• patch-mosaic
• savanna fires
• seasonal-mosaic

关键词:

• 非洲
• 焚烧制度
• 补丁马赛克
• 草原火灾
• 季节性马赛克

Palabras clave:

• África
• régimen de incendios
• mosaico de retazos
• incendios sabaneros
• mosaico estacional

Acknowledgments

The author would like to thank Chris Duvall, the editor, and six anonymous reviewers for their helpful comments, as well as Fakuru Camara for his help in the field, and Philip Laris for his careful editing of the text.

Notes

1. I use the term mesic to distinguish savannas with rainfall levels between 800 and 1,100 mm per year following Menaut, Lepage, and Abbadie (1995).

2. See Biggs et al. (2003) and Verlinden and Laamanen (2006) for exceptions.

3. The length of time it takes for a burn scar to fade and become undetectable varies significantly depending on the data and microenvironmental factors. Using the burn scar detection method described earlier, I was able to distinguish burned areas that were up to two months old in many cases. Older burn scars appeared faded or “lighter” in the composite images due to the fact that char and ash had been removed by wind.

4. My use of the MSD is somewhat different from that most commonly found in the remote sensing literature. The MSD is most often used spatially to describe the variation between pixels across space (usually a moving window), but it is sometimes used temporally to determine the variation over time when pixels are being composited (such as with coarse resolution data, which has high temporal frequency).

5. The percentages given do not include the unclassified pixels, which made up about 15 percent of the image. Unclassified pixels were those that changed cover type during the study period (for example, falling water levels would result in different classifications for river pixels) or confused pixels that were left unclassified by the algorithm. Water bodies appear slightly large in the image due to the masking process that included the high-water marks.

6. Modified from Laris (2006).

7. Results are based on those 41 percent of the respondents who admitted they set fires. The differences in the responses of those who stated they set fires and those who did not are reported in Laris (2002).

8. The Hunter Society (Donson Ton) is a pan-national, panethnic organization composed of village-level units. Traditionally, the society controlled access to, and use of, areas outside of the immediate (farmed) area of a village (Cisse 1994).

9. Le Page and colleagues (2010) used active fire data, which are distinguished from burned area maps in that they use the thermal band of a remote sensor (usually at the 1 km scale) to detect flames. Because very small fires can be detected, these data are useful for documenting the timing of fires but not necessarily the extent of the burned area.

10. Menaut et al. (1991) found that 25 percent to 50 percent of the Sudan savanna and 60 percent to 80 percent of the Guinean savanna in West Africa burned annually. Eva and Lambin (1998) found slightly lower values of 28.2 percent for the Sudan zone and 51.8 percent of the Guinea zone of central Africa. Barbosa and colleagues (1999) found higher levels of burning in the Sudan zone (70.1 percent) and lower values in the Guinea zone (57.7 percent). Devineau and colleagues (2010) found that 42 percent to 74 percent of their study area in neighboring Burkina Faso burned.