Large isolated trees are keystone structures that can help maintain biodiversity in fragmented landscapes, with evidence that open areas with isolated trees may support similar levels of taxonomic diversity to nearby patches of habitat. However, it is not clear if isolated trees can support the same diversity of ecological functions as trees in habitat patches. We compared species richness, community composition and functional diversity of birds in trees at forest edges and isolated trees. Twenty isolated trees and ten edge trees of American muskwood (Guarea guidonia) were selected, and sampled on 11 occasions, each of 20 minutes, over four months. All individual birds that landed at trees were recorded using a standardized protocol. Species richness was, on average, almost twice as high at edge trees than at isolated trees. Taxonomic composition differed between edge and isolated trees, with many forest-dependent birds restricted to edge trees, and some open-area birds restricted to isolated trees. Overall functional diversity was similar at edge and isolated trees, but some ecological functions (e.g. frugivory) were less frequent, while others (e.g. granivory) were more frequent at isolated compared to edge trees. Isolated trees are important for supporting many ecological functions in modified areas. However, the maintenance of forest patches is essential to complement the provision of such functions in modified landscapes.

We conducted this study in forests and pastures located in a countryside of Rio de Janeiro state, Brazil (within and adjacent to the Reserva Ecológica de Guapiaçu; 22°24’S, 42°44’W). We carefully selected trees for sampling, controlling possible confounding factors. To control for potential differences in the attractiveness of different tree species to birds, we focused on a single tree species, American muskwood (G. guidonia), which is the dominant isolated tree in the study area (A. D. Azevedo, unpublished data) and common throughout the Atlantic Forest (Cartes, 2003; Lima et al., 2009). We measured and georeferenced 102 G. guidonia trees within the study landscape and selected 20 isolated trees and 10 edge trees based on a careful systematic procedure. First, we selected trees that were at least 200 m apart, to prevent double counting of birds and to maximize the independence of sampling units (trees). Second, we selected trees spanning a broad range of sizes (i.e. circumference at breast height). Third, isolated trees varied in the number of surrounding isolated trees (from all species) and in the amount of surrounding forest cover, thus covering the range of conditions present within our study landscape. Edge trees were located at the edge of a large continuous forest area of ~ 100,000-ha. Thus, the edge trees served as a reference to determine the potential pool of bird species present at the edges of continuous forest that might use the sampled isolated trees. Fruit density of all isolated and edge, i.e., the volume of the canopy with fruits estimated by counting and attributing a score from 0 to 5.

We sampled birds from September to December 2016, which corresponded to the period of greatest reproductive activity (Sick, 1997). Based on a pilot study conducted in September 2015 and information in the literature (Fischer and Lindenmayer, 2002a, b; DeMars, Rosenberg & Fontaine, 2010), we determined 20 minutes to be the appropriate period for sampling. During each 20-min sampling event, an observer stood at 5 m from a sampled tree (ideal distance for bird identification) and recorded all birds that landed at the tree. Surveys were conducted between 5:00 and 11:00 a.m. and only during fine weather, with the same amount of effort employed for all trees. All trees were sampled in each of the four months of the study at all survey hours, summing to 11 sampling events of 20 min each (220 min in total for each tree).

To quantify functional diversity, we selected bird traits related to habitat requirements (open area, generalist, or forest; Stotz, 1996; del Hoyo et al., 2020), diet (carnivore, frugivore, nectarivore, granivore, insectivore, saprophage, or omnivore; Wilman et al., 2014), foraging strata (terrestrial, understory, midstory, superior, or generalized; Stotz 1996, Wilman et al., 2014) and body mass (small – individuals smaller than the median, i.e. 24.75 g, and large – individuals larger than the median; Dunning, 2007), which are related to both the response of species to environmental change and their effects on ecosystem function (Luck et al., 2012). Therefore, we compiled data for a total of four traits containing 16 categories. The traits classification was based on available information in the literature (e.g. Stotz, 1996; Dunning, 2007; Wilman et al., 2014; del Hoyo et al., 2020).

Three documents are available: one containing the information of each sampled tree (trees.csv); another containing the recorded species in the trees, and the number of times each species was recorded (bird composition.xlsx); and the last one containing the traits of each species (traits.xlsx).