Skip common site navigation and headers
United States Environmental Protection Agency
Great Lakes Ecosystems
Begin Hierarchical Links EPA Home > Great Lakes EcosystemsUpland Ecosystems > 1994 Oak Savanna Conferences > William A. Hoffmann
Aquatic Ecosystems
EPA Region 5 Critical Ecosystems
Ecosystem Funding
Great Lakes Basin Ecosystem
Great Lakes Biological Diversity
Green Landscaping
Rivers and Streams
Upland Ecosystem

1994 Proceedings
North American Conference on Savannas and Barrens


William A. Hoffmann
Department of Organismic and Evolutionary Biology
Harvard University

Living in the Edge: 1994 Midwest Oak Savanna Conferences

The Cerrado region of Brazil occupies an area of approximately 2,000,000 km2, or roughly one fourth the area of Brazil (Adamoli et al. 1985). The region is dominated by savanna vegetation ranging in density from open grassland with no woody cover to closed woodland with a sparse grass layer. The region also includes areas of seasonal forest and gallery forest, but these are not normally considered to be Cerrado vegetation.

Environmental factors long considered to be important in the ecology of the Cerrado are water, nutrient status, and fire. While most of the Cerrado region receives an annual precipitation of 1200-1800 mm that is unevenly distributed throughout the year. Most of the region has a 5-6 month water deficit (Adamoli et al. 1985). The characteristics of the soil of the Cerrado region also pose a serious problem for plant growth. A combination of low pH, high aluminum saturation, and low concentrations of P, K, Ca, Mg, Zn, and Cu in Cerrado soils (Lopes and Cox 1977) is severely limiting to plant growth (Haridasan 1992). The third important factor, fire, is a widespread and ever-increasing factor in the Cerrado. While fire has probably long been an important factor in the Cerrado, fire frequency is undoubtedly much higher in historic times. In areas of Cerrado inhabited by humans, fire intervals of two or three years are common.

The Cerrado vegetation has a great capacity to regenerate following burning. Many tree species are protected by a thick layer of bark, and others resprout quickly following fire. The apparent ability of the Cerrado to return to the original structure following a burn gives the impression that the Cerrado is little affected by fire. For example, Eiten (1984) states that fire has little affect on the tree layer and only a temporary affect on the shrub layer. However, Ramos (1990) has shown that repeated burning on a biennial basis results in a tree basal area one half that of areas protected for 13 years. Studies in the llanos of Venezuela have shown an increase in tree species richness during a 25 year span of fire protection (San Jose and Farinas 1991). The long-term effects of fire are not the same for all species. Moreira (1992) has shown that some species of woody plants are more abundant in frequently burned sites, while others are more abundant in fire-protected sites.

Little can be said about the how these long-term changes come about. Little is known about the population dynamics of woody plants in the Cerrado, and even less about the effect of fire on population dynamics. One important, yet unresolved question of the population biology of Cerrado woody plants is the relative importance of sexual verses vegetative reproduction. Early workers noted an apparent scarcity of seedlings and an ability of many species to reproduce sexually via suckering (Ferri 1961; Rizzini and Heringer 1962). Likewise it has been assumed that the long dry season is too harsh to permit survival of seedlings which have not yet developed a root system extensive enough to reach the soil layers where moisture is available during the dry season. But recently, it has been shown for two tree species that survival of seedlings is quite high even during the first dry season (Oliveira and Silva 1993).

In order to study the role of fire on the population dynamics of woody plants, I have chosen a comparative approach. Species choice was guided by overall abundance at the site and the effect of repeated burns on abundance. Six species of woody plants were chosen for study: Rapanea guianensis (Myrsinaceae), Miconia albicans (Melatastomataceae), Periandra mediterranea (Fabaceae), Rourea induta (Connaraceae), Piptocarpha rotundifolia (Compositaceae), and Roupala montana (Proteaceae). The first three species have been shown to be less common in frequently burned sites as opposed to sites protected from fire (Moreira 1992). Roura induta and P. rotundifolia are more common in frequently burned sites (Moreira 1992). Roupala montana shows a reduced abundance in frequently burned sites (Moreira 1992) but is still quite abundant at these sites (personal observation). Approximate maximum heights of these species are: R. guianensis, 9 m; M. albicans, 4 m; P. mediterranea, 2.5 m; R. induta, 5 m; P. rotundifoilia, 6 m; R. montana, 7 m.


This study is being conducted in two adjacent ecological reserves, IBGE (Instituto Brasileiro de Geografia e Estatistica) and JBB (Jardim Botanico de Brasilia). The site lies with in the Federal District of Brasilia at 15_52'S and 47_51'W. The soil is a well-drained oxisol with depths of up to 30 m. Annual rainfall is approximately 1500 mm, falling mostly within a well-defined wet season from October to April.

Located at the site is a large project established to study the effects of fire in the Cerrado. A total of 30 experimental plots, each 10 ha, have been established across a physiognomic gradient from open shrub savanna to closed savanna woodland. Each plot belongs to one of five fire regimes: control, biannual early-season burns, biannual mid-season burns, biannual late-season burns, or quadrennial mid-season burns. Thus each fire regime is represented by six replicates, two in each of three physiognomic categories.


Twelve plots were established according to a 2 x 3 factorial design. The first factor is fire treatment (control and late-season biannual burns) and the second factor is physiognomy (open shrub savanna, 'typical' Cerrado, and dense Cerrado). Each combination of levels is represented by two replicates. Of the burn treatment plots, three were burned in September, 1992 (one of each physiognomy), and three were burned in September or October, 1993.

The standard size for these plots is 15 m x 25 m. Within the center 5 m x 15 m 'core' of each plot, every individual of the six study species was tagged and mapped. A thorough search was made to include even the smallest individuals. Yearly surveys of these cores areas are conducted to follow reproduction via production of new suckers In the remaining area of the plot the individuals of the study species are tagged and mapped, but less care was taken to include all individuals of the smallest size classes. Large individuals lying outside the plot were also included to increase the sample of individuals in the largest size classes, which were underrepresented in the plots.

The following measurements were taken yearly on all individuals: Height, length (defined as the distance to the furthest apical bud from the point of insertion of the main stem in the soil), crown diameter, and stem diameter. These measurements were taken in the dry season when little growth is occurring.

Censuses to estimate seed production were conducted in the plots in the appropriate time of year, depending on the species in fruit. In addition to censuses conducted on the marked individuals, transects were walked to sample individuals in plots of the other fire regimes of the main fire project in order to estimate reproduction at various times since burning.

Field germination experiments were conducted to test the effects of woody cover and time since burn on the recruitment of seedlings. A 3 x 4 factorial design was utilized. Three densities of woody cover (1. No woody cover, 2. Medium cover, offering approximately 30-50 % cover, 3. dense cover, offering approximately >75 % cover) and four times since last burn (approximately 0-3 months, 1 year, 2 years, or >7 years) were used. A total of 3 replicate plots were established for each combination of density and time since burning. This was carried out for five of the principal species and an additional 7 species for which sufficient seeds could be collected. The number of seeds per plot differed between species, depending on the availability of seeds.



All species showed low mortality rates in response to burning. Mortality in burned plots of individuals of reproductive size were less than 6% for all species.

Vegetative Reproduction

Two species, M. albicans and P. mediterranea, showed no evidence of ability to reproduce vegetatively. A third species, P. rotundifolia, produced some suckers, but the overall density of this species in the study plots was too low to make any meaningful comparison. Three species showed an increase in sucker production in burned sites. By the end of the first growing season following burning, R. guianensis produced an average of 1.40 suckers/ individual in burned sites as opposed to 0.18 suckers/individual in unburned sites. Rourea induta produced 0.14 suckers in burned plots and 0.01 suckers/individual in unburned plots. Roupala montana produced 0.47 suckers/individual in burned plots and 0.14 suckers/ individual in unburned plots. These values represent the ratio of new suckers produced to the initial number of individuals in the core areas of the study plots.

Seed Production

The effect of burning on seed production differed between species. Three species, M. albicans, P. mediterranea, and R. induta, produced no seeds in plots burned one year previously but seed production returned to control levels in plots burned two or three years previously. Rapanea guianensis also showed a drastic decline in reproduction in the first year following burning, but even in plots burned three years previously, seed production is considerably lower than control levels. Piptocarpha rotundifolia shows a doubling in seed production in the year of burning, returning to control levels in subsequent years. At the time of writing, comparable data had not yet been collected for R. montana, but it appears that it will exhibit a behavior similar to that of R. guianensis.

Seedling Establishment

Of the twelve species which were included in the field germination experiment, five showed significant effects of time since burning, and seven showed significant effects of density of woody cover. For all species in which time since burning had a significant effect, establishment was lower in recently burned plots than in plots not recently burned. No species showed increased establishment in burned sites. Of the seven species exhibiting a significant response to density of woody cover, six had higher establishment in closed sites than open sites. Only M. albicans had higher success in open sites.


My results shows a considerable effect of fire on the population biology of woody plants of the Cerrado by reducing the importance of sexual reproduction relative to vegetative reproduction. This shift in importance comes about through four processes:1) increase in production of suckers following burning; 2) decrease in seed production for many species of woody plants; 3) decrease in seedling establishment following burning; and 4) decrease in woody cover which is facilitative for the establishment of seedlings.

Increase in sucker production following fire is likely a response to destruction of the above-ground portion of individuals rather than a direct response to fire. Stoeckeler and Mason (1956) have shown an increase in sucker production of aspen in cut stands verses undisturbed stands. Likewise, burning has been found to increase sucker production in some Australian savanna species (Lacey et al 1982). Initiation of root buds is inhibited by auxins and stimulated by cytokinins (Peterson 1975). Removal of the aerial parts of a plant results in an decrease in auxin relative to cytokinin, likely causing an increase in sucker production.

Fire reduced seed production in five of the six principal study species. This seems to be a typical response of Cerrado trees and shrubs to burning. While burning stimulates flowering among many herbaceous and subshrub species, it commonly reduces flowering among trees (Coutinho 1976). Likewise, fire tends to reduce the size of many individuals, which must certainly reduce the reproductive output in subsequent years.

My results show a reduction in seedling recruitment in recently burned sites. This is contrary to results often reported for vegetation in which seedling establishment is inhibited by the deep shade or thick litter found under a mature canopy. In the Cerrado, such conditions may be found in some of the more closed sites but it does not seem to be the case in typical Cerrado. It is more likely that the thin layer of litter present in unburned sites is facilitative to seedling establishment. The loss of this layer, along with the loss of cover provided by standing dead biomass of grasses, probably contributes to the reduction of seedling establishment in burned sites.

The facilitative effect of woody cover on establishment and growth of seedlings shown here has been documented for other ecosystems (Kellman 1985; Lonsdale and Abrecht 1989; Bowman and Panton 1993). The immediate reduction in cover due to burning and the long term reduction in tree density resulting from repeated burns should result in an overall reduction in seedling establishment.

The reduction in importance of sexual verses vegetative reproduction following burning is certain to effect both population and community dynamics. It can be expected that fire will have a greater negative effect on species incapable of reproducing asexually. Indeed, P. mediterranea and M. albicans, both of which reproduce exclusively by seed, are species which are less abundant in frequently burned areas (Moreira 1992). Species which are capable of reproducing both sexually and asexually probably exhibit a shift in reproduction from sexual to asexual as frequent fire becomes a factor.


Adamoli,J., J. Macedo, L. G. de Azevedo, J. M. Netto. 1985. Caracterizacao da regiao dos Cerrados. In W.J. Goedert, editor. Solos dos Cerrados. EMBRAPA: Brasilia.

Bowman D. M. J. S. and W. J. Panton. 1993. Factors that control monsoon-rainforest seedling establishment and growth in north Australian Eucalyptus savanna. Journal of Ecology 81:297-304.

Coutinho, L. M. 1976. Contribucao ao Conhecimento do Papel Ecologico das Queimadas na Floracao de Especies do Cerrado. Ph.D. Thesis. Universidade de Sao Paulo.

Eiten, G. 1984. Vegetation of Brasilia. Phytocoenologia 12:271-292.

Ferri, M. G. 1961. Aspects of the soil-water-plant relationships in connection with some Brazilian types of vegetation. In: Tropical Soils and Vegetation. UNESCO.

Haridasan, M. 1992. Estresse Nutricional. In: B.F.S Dias, ed. Alternativas de Desenvolvimento dos Cerrados. IBAMA: Brasilia.

Kellman, M. 1985. Forest seedling establishment in Neotropical savannas: transplant experiments with Xylopia fructescens and Callophyllum brasiliense. Journal of Biogeography 12:373-379.

Lacey, C. J., J. Walker, and I. R. Noble. 1982. Fire in Australian tropical savannas. In: B.J. Hunter and B.H. Walker, eds. Ecology of Tropical Savannas. Springer-Verlag: Berlin.

Lonsdale, W. M. and D. G.Abrecht. 1989. Seedling mortality in Mimosa pigra, an invasive tropical shrub. Journal of Ecology 77:371-385.

Lopes, A. S., and F. R. Cox. 1977. A survey of the fertility status of surface soils under "cerrado" vegetation in Brazil. Soil Science Society of America Journal 41:742- 747.

Moreira, A. G. 1992. Fire Protection and Vegetation Dynamics in the Brazilian Cerrado. Ph.D. Thesis. Harvard University: Cambridge.

Oliveira, P. E. and J. C. S. Silva. 1993. Reproductive biology of two species of Kielmeyera (Guttiferae) in the Cerrados of Central Brazil. Journal of Tropical Ecology 9:67-79.

Peterson, R. L. 1975. The initiation and development of root buds. In: J. G. Torrey and D. T. Clarkson, eds. The Development and Function of Roots. Academic Press.

Ramos, A. E. 1990. O Efeito da Queima sobre a Vegetacao Lenhosa do Cerrado. M.S. thesis. Universidade de Brasilia: Brasilia.

Rizzini, C. T. and E. P. Heringer. 1962. Studies on the underground organs of trees and shrubs from some Southern Brazilian Savannas. Anais da Acadamia Brasileira de Ciencias 34:235-247.

San Jose, J. J. and M. R. Farinas. 1991. Temporal changes in the structure of a Trachypogon savanna protected for 25 years. Acta Oecologia 12:237-247.

Stoeckeler, J. H. and J. W. Mason. 1956. Regeneration of aspen cutover areas in northern Wisconsin. Journal of Forestry 13-16.


Begin Site Footer

EPA Home | Privacy and Security Notice | Contact Us