Abstract

<jats:title>ABSTRACT</jats:title><jats:p>Seeds of the Brazilnut tree (<jats:italic>Bertholletia excelsa</jats:italic> Humb. &amp; Bonpl., Lecythidaceae) sustain one of the most important extractive industries in neotropical forests. Yet little is known about the demography and seed dispersal ecology of <jats:italic>Bertholletia</jats:italic>, particularly in natural stands which have not been previously harvested. This study presents data on the population density, spatial distribution, and seed dispersal ecology of Brazilnut trees at a pristine stand located within the Kayapó Indian Area of southeastern Amazonia, Pará, Brazil. Brazilnut trees were primarily found within groves (<jats:italic>castanhais</jats:italic>) of 75 to 149 trees, with a few isolated trees in between. Although the density of trees ≥ 10 cm in diameter at breast height (hereafter, dbh) at two groves was 4.8 to 5.1 trees ha<jats:sup>–1</jats:sup>, the overall density for the entire study area of <jats:italic>c.</jats:italic> 950 ha was estimated at 1.3 tree ha<jats:sup>–1</jats:sup>. Within-grove nearest neighbour distances averaged 21 m and were markedly skewed towards even shorter distances. Seed dispersal experiments using 709 marked seeds indicated that this pattern can be largely explained by the highly restricted seed shadows imparted by the main seed dispersal agents of <jats:italic>Bertholletia</jats:italic> at this site, the red-rumped agouti (<jats:italic>Dasyprocta leporina</jats:italic>). Agoutis on average scatterhoard <jats:italic>Bertholletia</jats:italic> seeds to an average distance of 5 m, and rarely beyond 20 m, from seed stations. We suggest that, once edaphic and climatic conditions are suitable, the highly contagious spatial distribution of <jats:italic>Bertholletia</jats:italic> trees at the landscape level can be largely accounted for by the quantitatively dominant effect of short-distance dispersal by caviomorph rodents, and rare events of long-distance dispersal provided by other vectors. This mechanism of grove formation need not resort to untested conjectures of human dispersal and intentional planting in prehistoric and historic times as it has often been suggested in the literature.</jats:p>