CANCELLED - Use of canopy gaps by birds in the central Amazon: A quantitative approach with imperfect detection

Background/Question/Methods

The vastness of Amazon forests creates a false appearance of homogeneity; however, forest organisms do not occur with the same probability in every location. Tree-fall gaps, in dynamic succession from tree fall to complete gap disappearance, create spatial variation in vegetation density and are a strong source of habitat heterogeneity inside the forest. Aiming to understand effects of habitat heterogeneity on habitat use and local distribution of bird species, we investigated differences in site-occupancy by birds in closed forest and in the vicinity of tree-fall gaps. We tested predictions of different occupancy for 68 bird species of terra firme forest of the Amazon Guianan shield, north of Manaus, Brazil. Specifically, we tested how occupancy of each habitat type should vary among three pre-determined species categories of habitat use. We used autonomous recording devices to collect data in 57 closed forest and 50 tree fall gap sites from June to October 2010. Our analytic approach, a multi-species hierarchical community model that accounts for the possibility of detection failure, leads to strong inferences about changes in occupancy between habitat types, species and groups of species.

Results/Conclusions

In the broadest sense, our results did not confirm predicted diferences in occupancy across groups. In general hypothetical ‘gap-seeking’, ‘gap-avoiding’ and ‘neutral’ groups had similar occupancy probabilities within the same habitat; only one group had different occupancy between habitat types. At the species-level, only 25% of investigated species conformed to predicted occupancy differences between habitat types. Hypothetical ‘gap-seekers’ had the highest detection probabilities in both habitats. Interestingly, hypothetical ‘gap-avoiders’ had higher probability of detection on gap areas than in closed-forest areas. ‘Neutral’ species, those predicted to occupy any habitat type with the same probability, had the highest detection probabilities in closed forest across groups. In general, the ornithological classification of bird habitat preferences used to form our occupancy predictions did not match our field estimates. These results highlight the need to test widely accepted classifications of habitat preference as well as the usefulness of hierarchical community models for testing hypotheses about groups of species.