Site

Liz, Júlia Alberti de, Mikich, Alessandra, Macedo, Vitória Ramos, Gonçalves, Rodrigo Barbosa & Mikich, Sandra Bos, 2024, Spatial gradient and vegetation cover affecting species diversity and composition of bees in Atlantic Forest remnants of South Brazil, Revista Brasileira de Entomologia (e 20240056) 68 (4), pp. 1-12 : 3

Site comparison

We compiled the data from the ten RMC bee assemblages sampled by Graf et al. (2022) to contextualize the Colombo assemblage ( Table 1). Graf et al. (2022) sampling design can be considered similar, each site was sampled monthly over a year, using the same methodology described above. For the Colombo dataset, the species Apis mellifera was excluded from comparative analyses since it was not sampled in the other sites. The sampled location here is designated as Site 11 to maintain the original site numbering.

Comparative analyses were conducted in R version 4.2.3 through RStudio (RStudio Team, 2020; R Core Team, 2021). To estimate diversity, Hill Numbers were calculated using the “iNEXT” package (version 3.0.0) (Hsieh aet al., 2022), which generates rarefaction and extrapolation curves for species diversity. The first three Hill numbers were used: q = 0 (species richness), q = 1 (the exponential of Shannon’s entropy index), and q = 2 (the inverse of Simpson’s concentration index) ( Chao et al., 2014). For extrapolation, the established endpoint (1400) was approximately double the lowest recorded abundance, 701 individuals at Site 10. The confidence interval curves were used as a test to detect differences in richness and diversity among areas. We also estimated species diversity (Hill numbers with q = 0, 1 and 2) with a subset of 700 individuals to evaluate their responses to the predictors described below. Site ordering based on species composition was performed using the “metaMDS” function from the “vegan” package ( Oksanen et al., 2022), with a species abundance matrix for each site. Non-Metric Multidimensional Scaling (NMDS) was chosen as it is an ordination technique that simplifies the analysis of multivariate data, organizing sites according to environmental variables and species compositions ( Gotelli and Ellison, 2004).

To investigate the factors that might influence species composition, we chose an approach that used geospatial information from the sampled areas, specifically the north–south and east–west spatial gradients.These predictors were represented by the latitude and longitude coordinates of the 11 sampling sites. In particular, the north–south gradient may represent both the original RMC vegetation type, i.e. the tendency of Mixed Ombrophilous Forest areas on the Northern region and Natural Grasslands on the Southern region of RMC, as well as larger vegetation fragments occurring in the north ( Maack, 1931; Klein and Hatschbach, 1962). The Mantel test was conducted to verify the existence of a spatial correlation between species composition and spatial gradients, i.e., to check if closer locations have more similar species compositions due solely to their proximity.

We also quantified the vegetation cover, that being the percentage of land that is covered by the existing vegetation within a 1 km buffer of sampling site centroid. A simple regression was performed for 700 individual diversity estimators and the three predictors. To evaluate the beta diversity response, we used the “envfit” function to project vectors into the ordination space representing the maximum correlation of predictor variables with the assemblages. The significance of each variable was tested by permutation test, with vectors analyzed separately, returning r 2 and p values that define the explanatory capacity of variables regarding the species composition at each location.