Grounded ground birds and surfing canopy birds: Variation of foraging stratum breadth observed in neotropical forest birds and tested with simulation models using boundary constraints

Vertical stratification of bird species has been intensively studied in temperate as well as tropical forest bird communities. One question that has received less attention is the variation of foraging stratum breadth. Hypotheses trying to explain that variation are based on differences in perception, evasion, adaptation, competition, and environmental and geometric boundary constraints. Although some predict that foraging strata near the ground should be narrower than those near the canopy, others predict that strata in the midstory should be broadest. These hypotheses were tested using observational data of 92 rainforest bird species. Observations were made from or near a 40 m high canopy crane situated close to the Surumoni River (38109N, 658409W) near Esmeralda, Estado Amazonas, southern Venezuela (Surumoni-Crane-Project). I found a highly significant parabolic relationship between mean foraging height and its standard deviation showing that midstory bird species foraged in a broader stratum than either understory or canopy species. Using a conservative comparative method, I also showed that the relationship was not the result of the phylogenetic relatedness of the species involved. Analysis of previously published data supports the relationship for two other tropical avifaunas. Because the ground and the outer canopy represent geometric boundaries to themovement ofmost forest birds, two simulationmodels were used to test whether midstory species may simply have a broader stratum because they are less constrained in their movements than species near the boundaries. The ‘‘random model’’ allowed individuals to move randomly between the two boundaries. It illustrated that boundary constraints alone cannot explain the observed parabolic relationship. In the ‘‘preference model,’’ individuals tended to move back to their preferred height within the forest column. That model simulated the observed parabolic relationship once a sufficient number of time steps had been reached, suggesting that a simple combination of geometric boundaries and preference for a specific foraging height may generate narrower strata near the boundaries. The other hypotheses only explain the parabolic relationship if other factors (e.g. vegetation density) are vertically distributed in such a way as to generate the observed parabolic pattern. Finally, I briefly discuss the advantages and disadvantages of using raw data in across-species analyses versus corrected data in phylogenetically-controlled analyses.