CHAPTER 3 - VEGETATION STRUCTURE AS AN INDICATOR FOR LAND-COVER DYNAMICS ASSESSMENT IN THE AMAZON

'Esse mato tem palmeiras…'

3.1. Disturbance and secondary succession in the Amazon: short review and motivation

The ecological literature on succession is extensive. Rather than making too broad a review about the topic, the focus is on some concepts related to vegetation regrowth as a background for this study in tropical forests of Rondônia. Since Cowles' observations around the shores of Lake Michigan (Cowles 1911) and Clements'(1916, 1928, 1936) numerous studies, many theoretical ideas about succession in several fields have been argued. According to Drury and Nisbet (1973), in spite of all the different approaches and competing conceptualizations, there is considerable agreement on the general trend of succession. During the 1960s, authors attempted to develop a general theory of structural and functional characteristics of a community's development (Hutchinson 1965, MacArthur et al. 1966, Levins 1968). Odum (1969) proposed a model for the process, discussing how the energy balance in the ecosystem progressively changed. Whittaker (1970) described changes in different vegetation variables through the course of succession ending up in a climax community. Most of these studies focused on temperate regions. The application of the succession concept to tropical forest ecosystems started gaining importance when Richards (1952) called attention to processes following disturbance. Since then, many studies have been carried out in different tropical sites. The paragraphs below describe some findings of works completed in the 1980s and 1990s.

Forest communities show variation due to availability of flora, biophysical differences between formations, and disturbances. Within formations, variation is related to topography, soils, seedling arrival, and success (Whitmore 1998). The fragility or susceptibility to damage from disturbance in a forest ecosystem is a consequence of its ecological characteristics. In tropical rain forests, most soils are infertile, with a low content of nutrients, making the nutrient cycling an important mechanism to maintain the ecosystem. When the process is disturbed, nutrients can be rapidly lost (Jordan 1989).

There are various sorts of forest disturbance. The greater the disturbance of a mature forest the longer it will take to recover. The extreme situation is when disturbance produces intense degradation with no chance for recovery. In this case, the process may be followed by ecosystem degradation (loss of structural and functional integrity), environmental degradation (loss of populations or critical functions), biodiversity degradation (loss of genetic diversity), and agricultural degradation (loss of productivity) (Vieira et al. 1993). Degradation may increase environmental risks such as flammability (Nepstad et al. Flames 1999). But when these extremes do not occur, succession starts the recovery of vegetation in a dynamic process. Abandonment brings rapid transformation to a competitive environment that induces successional change (Kellman 1980).

Secondary tropical forests originate from some source of disturbance (Corlett 1995). Succession generally refers to changes in species composition and abundance during or following disturbance of a site. The process is dependent on four main sources of recovery: regeneration of remnant individuals, germination from the soil seed bank, sprouting from cut or crushed roots and stems, and seed dispersion and migration from other areas (Tucker et al. 1998).

Sharp distinctions between successional stages are often artificial (McCook 1994), but useful to differentiate between forest or secondary formations. The most common distinction within forest species is between two contrasting ecological groups: pioneers (short- and long-lived) and climax species (Whitmore 1998). Climax species can germinate and establish seedlings below a canopy, whereas pioneer species require full light. Therefore, succession is the process where pioneer (light-demanding) species establish themselves in big canopy gaps, climax (shade-tolerant) species follow, pioneers die creating small gaps, and mature forest species grow up.

The physiognomic outcome of this continuous process of restoration is a change in vegetation structure, analyzed in this chapter for the study area in Rondônia. Brown and Lugo (1990) enumerate five main structural characteristics typifying secondary forests: high total density but low density of trees > 10 cm diameter at breast height (DBH); low basal area; short trees with small diameters; low woody volume; and high leaf area indices. These characteristics change with time giving place to different stages of vegetation toward a forest formation.

Some succession trends are typical in tropical forests: initial floristic composition influences later stand composition; leaf area index (LAI) and production peak early in succession; and streamwater nutrient losses decline rapidly as biomass accumulates (Uhl and Jordan 1984). Successional vegetation appears to be better adapted than crop plants to the diminishing nutrient availability. Another important adaptation of successful successional species is their high dissemination capability and high sprouting capability after fire, both depending on disturbance intensity and duration (Vieira et al. 1996).

Several studies have been done to understand how characteristics of disturbance influence the rate of recovery. Although no one theory can explain all factors controlling succession, some variables appear to be more important than others. In the Amazon, three main factors control succession: availability of regeneration mechanisms (e.g., sprouts, seeds buried in the soil, seeds dispersed from surrounding areas); availability of seed germination and seedling establishment microhabitats (e.g., fruit trees and slash piles); and availability of nutrients, which may be affected by previous management (Uhl 1987). Tree species diversity and biomass accumulation vary depending on time and intensity of land use before recovering. A key factor retarding succession is the slow rate at which primary forest species become established on abandoned farms (Uhl et al. 1988). Some barriers to tree establishment include low propagule availability, seed predation, seedling predation, seasonal drought, and root competition with old vegetation (Nepstad et al. 1991). Although the natural forest recovery indicates a remarkable resilience, Saldarriaga et al. (1988) estimated that 190 years would be taken by a previously cultivated site to reach mature forest basal area and biomass values. Also, the number of tree species present after 40 years of succession is less than half the number in mature forests (Vieira et al. 1996).

In general terms, soil fertility and land-use history emerge as the critical factors influencing the rate of forest regrowth (Tucker et al. 1998). Uhl et al. (1982) found that the time of recovery depends on land use following removal. Logging, slash/burn/abandon, and slash/burn/agriculture/abandon cycles have increasing secondary succession duration. Large cleared patches, where seed sources are far away, may take hundreds of years to return to primary forest. In another study, Uhl et al. (1988) confirmed these findings. Different regeneration patterns occurred depending on land management following deforestation. Forest regenerated vigorously on sites of previously light use (biomass accumulation of 25% after 8 years.). Tree species richness was also high. Moderately grazed pastures also developed forest, but biomass accumulation and tree species richness were lower. Abandoned pastures subjected to heavy use had the least distinct patterns of succession (eight-year-old site was dominated by grasses and forbs). Only where land has been used too intensively for long periods is reforestation uncertain. Thus, in the absence of fire, forests recover on abandoned sites, accumulating biomass and species at a rate that is inversely related to the intensity of use prior to abandonment (Nepstad et al. 1991).

From the considerations above, it becomes clear that if we want to understand the variables affecting patterns of forest succession, we need to know the disturbance history. Recently, remote sensing and GIS have improved significantly the capability to monitor processes of LULC change (B. Turner 1995, National Research Council 1998). Land-cover classifications using these tools became fundamental to understand and monitor processes of deforestation and secondary succession, particularly in the tropics (Mausel et al. 1993; Moran et al. 1994, 1996; Foody et al. 1996; Steininger 1996). The integration of these methods of analysis, field data about vegetation structure and composition, and ecological research provide new opportunities for the study of dynamic processes such as forest disturbance and recovery at ecosystems and landscapes. For regional and landscape assessments, the study of vegetation structure in tropical forests is even more effective than floristic composition because of general spectral responses to vegetation communities at resolutions such as those in Landsat TM images.

In this chapter, the results for vegetation structure in Machadinho and Anari are presented as a basis for discussing the spectral response of secondary stages when using Landsat TM images. The rationale behind this approach is to follow an itinerary from the continuous vegetation variability found in the field to specific categories of secondary succession useful for LULC classifications such as presented in the next chapter.