Tropical rainforest are among the most complex and species rich-habitats on earth, harboring as much as two thirds of all the living animal and plant species (Beck, 2008; Waide, 2008). Numerous mechanisms that promote and maintain tropical species richness have been suggested e.g. pest pressure, specialization and disturbance (Leigh et al., 2004). The principal natural disturbance factor affecting the structure of rainforests are tree falls creating the gaps, which consist of an opening of the canopy that allows the sun light to reach the forest floor. Tree-fall gaps differ in their characteristics, like their size, depending on levels of intensity, frequency, extent and duration of the disturbances (Shea et al., 2004). The effect of the tree fall gaps is that tropical rainforests can be considered as a mosaic of micro-successional patches (Terborgh, 1992). The intermediate disturbance hypothesis proposes that the highest diversity is maintained at intermediate scales of disturbance (Connell, 1978). Empirical evidences that support this hypothesis have been reported for a wide range of species (Molino & Sabatier, 2001; Terborgh, 1992; Tomlinson, 1991; Hill et al., 2001).

Light is a major limiting factor in tropical forest (Chazdon et al., 1996), and the higher amount of light found in gaps leads to higher productivity compared with the closed forest (Denslow, 1987), playing an important role in plant species composition (Hubbell et al., 1999; Schnitzer & Bongers, 2002). Because of increased sun radiation some trees produce more fruits in the gaps than in the understory (Pinero, 1984; Denslow et al., 1986; Levey, 1990), and could attract more species e.g. birds (Levey, 1988), butterflies (Hill et al., 2001), and mammals (Beck, 2002; Beck et al., 2004). The great amount of resources provided by plant species growing in the gaps, compared to the understory, may favor many species feeding on them (Spitzer et al., 1997), e. g. butterflies.

Of the rainforest in the world, the Amazon rainforest is considered to host the greatest diversity of organisms (Olson & Dinerstein, 2002). The diversity found in the rainforests has been studied using assemblages of plants (Denslow, 1987), birds (Levey, 1988), mammals (Beck et al., 2004) and insects (DeVries et al., 1997; Hamer et al., 2003). Among these taxa, butterflies are often chosen as biological indicators (Lamas, 1997; Kremen, 1992; Hill et al., 2001; Fermon et al., 2005). Out of approximately 7000 species of butterflies found in the Neotropics, over 3500 occur in Peru (Lamas, 1997; DeVries, 1987). In Peru, where 60% of the territory is still covered with tropical rain-forest (Lamas, 1997), the national park of Manu offers a completely undisturbed ecosystem (Terborgh, 1992). The family Nymphalidae represents the most specious butterfly family, and they are also the largest family occurring in Peru (Lamas, 1997; Murray, 2000). Because of the butterflies distinct visibility (size and color) they are ideal models to address numerous ecological questions (e.g. DeVries, 1988; Hamer et al., 1997; DeVries et al., 1997, 1999; Shahabuddin & Terborgh, 2000). One widely used subgroup is the fruit-feeding nymphalids, which consist of species where adults feed on the juices of rotting fruit (DeVries et al., 1997, 1999).

Studies of butterflies conducted in rainforests have been focused on describing patterns of butterfly communities (DeVries et al., 1999; Fermon et al., 2005), using them as indicator species (Kremen, 1992), comparing faunas at different sites (Lamas, 1997), and impact of logging (Hamer et al., 2003; Hill et al., 2001; Fermon et al., 2005). Studies trying to describe the mechanisms affecting butterfly diversity in undisturbed rainforest had received less attention, with a few exceptions (Hill et al., 2001; Hamer et al., 2006).

In this study, we compared the assemblages of fruit-feeding Nymphalidae found in the undisturbed understory and natural tree-fall gaps of different sizes. Our hypothesis was that different nymphalids assemblages would occur in tree-fall gaps compared to the undisturbed understory. Furthermore because it has been shown that plant communities vary with gap size, we expect to find different densities and species by gap size.