Skip to main content

Coutinho et al. 2021 - Landscape structure is a major driver of bee functional diversity in crops

Cite this dataset

Coutinho, Jeferson Gabriel Da; Hipolito, Juliana (2021). Coutinho et al. 2021 - Landscape structure is a major driver of bee functional diversity in crops [Dataset]. Dryad.


The study of functional diversity can support the understanding of how changes at the landscape affects the roles of organisms in biological communities and consequently their ecological functions such as pollination. Different aspects of functional diversity have been addressed in recent years. Understanding those effects on landscape patterns is a key step to target ecological intensification. Still, elucidating those relationships requires studies in multiple spatial scales since effects and consequences are different considering biological groups and interactions. In that sense, by using a multitrait approach we evaluated whether the landscape structure and/or local environment characteristics could explain the functional richness, divergence and dispersion of bee communities in agroecosystems. In addition, we investigated to which extent  this approach help to predict effects on pollination services. This study was conducted in an agroecosystem situated in "Chapada Diamantina" region, State of Bahia, Brazil. Bees were collected using two complementary techniques (entomological nets and pan traps) in 27 sample units distributed orthogonally along a gradient of agriculture and landscape diversity. Bees were classified according to their response traits (e.g. body size, nesting location) and effect traits (e.g means of pollen transportation, specialty in obtaining some resource).The Akaike Information Criterion (AIC) was used to select the best models created through the additive combination of landscape descriptors (landscape diversity, mean patch shape and local vegetation structure) at the local, proximal and broad landscape levels. Our results indicate that both landscape heterogeneity and configuration matter in explaining the three properties of bee functional diversity. We indicate that different properties of functional diversity are influenced by different landscape descriptors, at different spatial scales. In this sense, including conjunct strategies, in different spatial scales, considering different ecological processes is a more relevant way to think about land use scenarios in a perspective to contemplate functional diversity.


Biological data were collected along a gradient of landscape proportion of agriculture and habitat types diversity at a 3 km scale. Two complementary sampling strategies were used: passive through the use of pan traps (Cane et al. 2000) and active, through entomological nets (Moreira et al. 2015). For the collection with pan traps, we used the three most common colors to attract bees:  yellow, white and blue. Blue and yellow traps also had ultraviolet radiation (Cane et al. 2000). In each pan trap was placed 120ml of water and approximately five drops of neutral detergent to break the surface tension.. The pan traps were installed at a height of 1m from the ground using 25mm diameter PVC pipes and perforated metal ribbons, used as support for 18mm diameter colored dishes. Traps were distant 5m from each other in an equidistant triangular shape, where each one was positioned at a vertex of this triangle. At each sampling point, three groups of three plates of different colors were installed, totaling nine pan traps for each sampling point (Figure 2). This set of nine pan traps formed a larger triangle in each sample unit. Each group of three was 15m from the nearest group.

The active sampling with entomological nets was made by two collectors walking through two isosceles triangles, within a reference hexagon with 25 m sides (Figure 3). All floral visitors seen on the flowers were collected with the nets between 07:30 and 17:30. For logistic reasons, the set of 30 sample units was divided into two groups of 15 along the study region. To avoid possible systematic effects of seasonality, each group was sampled every two months between January and November 2011, corresponding to two rainy seasons and two dry seasons. Each sample unit was sampled for 40 hours throughout the entire campaign. Collectors alternated the units they sampled, to minimize possible systematic effects on collection (Moreira et al. 2015). For the statistical analysis we used only 27 sample points, due to logistic difficulties associated with three of the sampling points during the collection period.

Usage notes



National Council for Scientific and Technological Development, Award: 556050/2009-6