Linking smallholder land use and fire activity: examining biomass burning in the Brazilian Lower Amazon

Abstract

Landscape forest flammability and regional biomass burning are critical environmental issues associated to fire use practices in the Brazilian Amazon. To better understand the gamut of these issues, all land use pathways that lead to fire use and culminate in different fire types must be investigated. An estimated 400,000 smallholders use fire in their land use practices, but most research has investigated connections between land use decisions and subsequent fire on large-scale cattle ranches only. This study presents research on smallholder fire use practices in the agricultural frontier south of Santarém, Pará. Vegetation vulnerable to smallholder fires is largely secondary succession, while only ∼8% is logged or mature forest cover. Fire efficiency in smallholder fire use practices is slightly higher than in previous studies used to estimate regional biomass burning, averaging 46.7–57.5%. In slashed succession areas, average efficiency also increases as biomass fuel load increases, but range increases as well, suggesting that smallholders may have more control over efficiency outcome in younger slashed succession areas than older slashed succession, logged areas or mature forests.

Introduction

Since the mid 1980s, unusually strong ENSO events have brought fire issues to the forefront in research on deforestation and tropical rain forests (Cochrane et al., 1999, Nepstadt et al., 1999). An estimated 10,000 km² of mature forest burned in Rondônia, Brazil, alone the later half of 1997 through 1998, while other large fires were detected in eastern Pará and the Tapajós Basin (Barbosa, 1998 sited in Cochrane and Sculze, 1998a). Although charcoal remnants suggest fire is a natural disturbance mechanism within tropical moist forest ecosystems, fire frequency is thought to be extremely low, at intervals in the hundreds if not thousands of years (Sanford et al., 1985, Saldarriaga et al., 1988, Turcq et al., 1998). Recent dramatic increases, however, are no doubt propelled by persisting development of the Amazon in combination with occasionally severe drought conditions. Field study evidence of potential forest damage from accidental fire is particularly startling. Whereas initial forest fires may cause some damage, an estimated 10% loss of live biomass, re-current fires within formerly burnt forest areas can destroy up to 80% of living biomass (Cochrane and Schulze, 1998b). Such fires may increase over all deforestation estimates in some regions by 129% (Cochrane et al., 1999).

Forest structure and conditions left after selective logging also bring concern of increased forest flammability as gaps in forest canopies and residual tree damage create dry fuel loads at forest ground level, that more easily combust during peak dry periods (Uhl and Buschbacher, 1985, Uhl et al., 1988, Fearnside, 1990, Uhl and Kauffman, 1990). The estimated 80–90% forest canopy of mature tropical moist forests can be reduced to 50% after experiencing logging (Nepstad et al., 1991). There is also evidence that burn or re-current burn damage in forests can vary by log extraction technique (Holdsworth and Uhl, 1997). The size of potentially vulnerable forest cover is significant: an estimated 10,000–15,000 km² of forest per year are currently damaged by logging crews (Instituto do Homen e Meio Ambiente da Amazônia, sited in Holdsworth and Uhl, 1997, Nepstadt et al., 1999). Yet, much of this research has been conducted on large-scale cattle ranches, which often include vulnerable logged forest patches. Less attention has been focused on links between those setting fires and other land use pathways which eventually culminate in fire activity, accidental fire and environmental degradation. An estimated 400,000 smallholders are distributed across Amazon territory and use fire in their land use practices (Serrão and Homma, 1993). Their potential impacts on local fire activity need further investigation.

In addition to field scale studies on forest flammability, biomass burning accumulation and potential trace gas emissions form a crux of regional scale research on fire issues in the Brazilian Amazon. Brazil’s contribution to global biomass burning has been estimated at 50–70% of total biomass burning (Myers, 1991), while annual CO₂ accumulation is calculated at 228 t × 10⁶ for the Brazilian Legal Amazon¹ in the base year 1990 (Fearnside, 1997). To derive regional biomass burning estimates, a multiscaled strategy is often employed which includes land cover classification of remotely sensed data and extrapolation of finer scale ecological processes (Houghton et al., 1991, Schroeder and Winjum, 1995, Fearnside, 1997). In particular, field survey data on biomass content within mature and secondary forest covers, estimates of fire efficiency, and calculations on replacement biomass, comprise the important finer scale variables necessary for estimation (Seiler and Crutzen, 1980, Brass et al., 1996, Fearnside and Guimaraes, 1996). Yet, error is often compounded through sampling scale or extrapolation, decreasing biomass burning estimation accuracy (Robinson, 1989, Rastetter et al., 1992, McGwire et al., 1993). For example, Malingreau and Belward (1992) found that burns can be under-or over-represented in remotely sensed data dependent on the spatial resolution of that data. While issues of scale translation in landscape ecosystem assessment and monitoring continue to be discussed (Turner et al., 1989, Levine, 1992, Rastetter et al., 1992), how accuracy is affected when translation involves both biotic (i.e., biomass content) and non-biotic processes (i.e., microclimatic conditions, human action, land area exposed to fire) has not been thoroughly addressed. To reduce potential inaccuracy, refinement of how local ecological variables interact with non-biotic processes, particularly human land use decisions is necessary (Sorrensen, 1998).

This paper examines smallholder fire use practices and presents a strategy to analyze ecological and non-biotic processes at the ground level and refine ecological variables used in regional biomass burning estimation. It presents a conceptual model to link biomass burning variables to land use pathways that influence these variables. Then it presents research focused on one specific pathway, that of smallholder farming. While multi-scale work is occurring, particularly on deforestation (Brondizio et al., 1994, Moran et al., 1994, Skole et al., 1994), this paper highlights those variables that are specifically important to biomass burning: biomass density, total area exposed to fire, burning efficiency, replacement vegetation, and vulnerability of vegetation to fire. Biomass burning is very much linked to deforestation, being the most common land clearance practice after trees have been felled. However, burning is also associated to a number of land management strategies after initial deforestation, making the temporal aspect of local fire use complex. Evaluating net emissions also depends on temporal understanding of land use and subsequent replacement vegetation (Fearnside and Guimarães, 1996). Concerns of global biomass burning merit strategies that cater specifically to fire use study and can be utilized in extrapolation to larger spatial scales.

The study region is located in the agricultural frontier south of the regional city of Santarém, Pará, and is contained in the municipality of Belterra (Fig. 1). Fire activity, forest flammability and field surveying for regional estimation has typically focused on the ‘arc’ of deforestation that runs through eastern Pará and along the southern flank of the Legal Amazon. Consequently, areas with a mixture of old spontaneous settlement and contemporary colonization have not been investigated. The study region is such an area and resides north central of this arc. Climate is tropical with an annual rainfall over 2000 mm and a distinct dry season of 2–3 months. Mean annual air temperature is 25°C. The native vegetation in the region is classified as terra firme, an upland forest cover that is not subject to periodic inundation by river systems (Olegário Pereira de Carvalho, 1992). These forests are usually situated on well-drained plateaus with heights exceeding 40 m. As human settlement has persisted, vegetation cover is more varied now. Up to 50 year old secondary succession from rubber plantations encircles the town of Belterra, while the long established rural area east of Belterra and the Santarém-Cuiabá Road (BR-163 in Fig. 1) contains largely younger secondary succession (less than 25 years old). South of Belterra, along the Santarém-Cuiabá Road smallholder lots still maintain between 50–70% primary forest, with the rest of land cover a mix of cropland, pasture and fallows (secondary succession).