Passive restoration of secondary forests can partially offset loss of biodiversity following tropical deforestation. Tree cavities, an essential resource for cavity-nesting birds, are usually associated with old forest. We investigated the restoration time for tree cavities suitable for cavity-nesting birds in secondary forest at the Biological Dynamics of Forest Fragments Project (BDFFP) in central Amazonian Brazil. We hypothesized that cavity abundance would increase with forest age, but more rapidly in areas exposed to cutting only, compared to areas where forest was cut and burned. We also hypothesized that cavities would be lower, smaller, and less variable in secondary forest than in old-growth forest, which at the BDFFP is part of a vast lowland forest with no recent history of human disturbance. We used pole-mounted cameras and tree-climbing to survey cavities in 39 plots (each 200 × 40 m) across old-growth forests and 11–34 year-old secondary forests. We used generalized linear models to examine how cavity supply was related to forest age and land-use history (cut only vs cut-and-burn), and principal components analysis to compare cavity characteristics between old-growth and secondary forest. Cavity availability increased with secondary forest age, regardless of land-use history, but the oldest secondary forest (31–34 years) still had fewer cavities (mean ± SE = 9.8 ± 2.2 cavities/ha) than old-growth forest (20.5 ± 4.2 cavities/ha). Moreover, secondary forests lacked cavities that were high and deep, with large entrances – characteristics likely to be important for many species of cavity-nesting birds. Several decades may be necessary to restore cavity supply in secondary Amazonian forests, especially for the largest birds (e.g, forest-falcons and parrots > 190 g). Retention of legacy trees as forest is cleared might help maintain a supply of cavities that could allow earlier recolonization by some species of cavity-nesting birds when cleared areas are abandoned.

We surveyed tree cavities in upland Terra Firme forest at the Biological Dynamics of Forest Fragments Project (BDFFP), north of Manaus in central Amazonas, Brazil (02^°20'S, 60^°5'W). 

At the time of our study, the area comprised a mosaic of old-growth forest, grasslands, and regenerating forest of different ages. Within this mosaic, we placed 39 plots: 8 in old-growth forest and 31 in secondary forest aged 11–34 years. Each plot measured 200 x 40 m (0.8 ha). For each stand age, we aimed to balance the number of plots in secondary forest that grew after cutting only vs. cutting and burning, but this was not always possible. Plots were centered along established linear transects or mist-net lines from other projects, with 200 m length and 20 m extension to each side. We spaced plots as well as was feasible given the existing trail network, which was limited, especially in secondary forest.

From July to October 2013, two observers searched for cavities by examining each tree within each plot, with the aid of binoculars. To inspect the interior of cavities, we used a 2-cm diameter video camera, which was either mounted on a 15-m telescoping fibreglass pole, or carried up to the cavity using single-rope tree-climbing (Politi et al. 2009, Di Sallo and Cockle 2022). Cavities were only included in our dataset if they had a sufficient diameter to permit the passage of a bird (>2 cm), and an internal chamber capable of offering protection to eggs (i.e., with a floor, walls, and some overhead protection, at minimum; Politi et al. 2009). We excluded cavities that contained water. We measured the height of each cavity from the lower sill of the entrance to the ground, using the 15 m pole, marked every 10 cm. For cavities above 15 m we used the climbing rope to measure cavity height. Entrance dimensions and horizontal depth were measured to the nearest centimeter using a tape measure (if we climbed to the cavity) or a 35 cm ruler attached to the top of the 15 m pole. We used a tape measure to determine vertical depth of cavities to the nearest centimeter. In the few cases in which it was not possible to climb to the cavities (e.g., dead trees in decomposition), but where we could access the cavity with the pole-mounted camera, we estimated vertical depth using the camera image, after first practicing on cavities that were measured with a tape measure. Using either tree climbing or the pole, we were able to measure 93% of cavities found in secondary forest and 92% of cavities found in primary forest.