Skip to main content
Dryad logo

Cross-scale drivers of woody plant species commonness and rarity in the Brazilian drylands


Pinho, Bruno X. et al. (2022), Cross-scale drivers of woody plant species commonness and rarity in the Brazilian drylands, Dryad, Dataset,


Aim: Locally abundant species are typically widespread, while locally scarce species are geographically restricted – the so-called abundance-occupancy relationships (AORs). AORs help explain the drivers of species rarity and community assembly, but little is known about how variation around such relationships is driven by species traits and niche-based processes, particularly in tropical woody plants. We tested the hypothesis that AORs in tropical dryland woody plants are positive and mediated by niche and functional traits along environmental gradients.

Location: The Caatinga dry forest and Cerrado savannah, Brazil.

Methods: We aggregated abundance and occurrence data into grid-cells representing local (10-km) to landscape scales (50-km). We calculated species mean relative abundance at occupied grid-cells (local abundance) and the proportion of grid-cells occupied (occupancy), and estimated their niche breadth and marginality along multivariate environmental gradients.

Results: AORs were positive but weak at different scales in both regions due to some locally abundant but geographically restricted species, with most species being both locally and geographically rare. Cross-species variation in local abundance was largely unpredictable, but occupancy was strongly driven by niche and functional traits, with a prominent negative effect of niche marginality. Geographically restricted species were associated with rare habitats, such as wetter and less intensively used habitats. Large seeds and abiotic dispersal favoured occupancy in Caatinga at small and large spatial scales, respectively, whereas species with conservative leaves were more widespread across scales in Cerrado.

Main conclusions: Woody plants in dry tropical biotas exhibit weak AORs, likely related to low habitat availability and dispersal limitation. Caatinga and Cerrado emerge as environmentally structured at multiple spatial scales, with several habitat-specialist rare species bearing specific regenerative and resource-use traits and relying on conditions threatened by climate change and land-use intensification. Examining AORs through the lens of niche, functional traits and spatial scales enables mapping patterns and drivers of species commonness and rarity, enhancing understanding of species assembly and providing tools for biodiversity conservation.


We first compiled a dataset of woody plant (i.e., sub-shrubs, shrubs, and trees) abundance within 330 plots, 121 across Caatinga dry forests and 209 across Cerrado savannahs (solid dots in Fig. 1).  Plots differed in size and stem diameter cut-offs (see Table S1, for further sampling details in each region), covering 122,326 plant individuals in Caatinga and 385,884 in Cerrado. Most of the plot data were extracted from the Neotropical Tree Communities database (TreeCo; Lima et al., 2020) plus our own data as described elsewhere (Ribeiro et al., 2015; Rito et al., 2017).

To assess macroecological patterns from relatively local to landscape scales, while also reducing pseudo-replication issues caused by nearby plots, we aggregated abundance data into spatial grid-cells of different sizes (10, 20, 30, 40, and 50 km) over the Caatinga and Cerrado regions (see an example of 50-km grid scale in Fig. 1). To avoid species with poorly known distributions/niches, only species occurring in at least three grid-cells were considered in further analyses. We also excluded palms, as they are frequently planted by locals in our focal regions. Finally, we filtered species for which functional trait data were available, which resulted in 135 and 417 woody plant species in the Caatinga and Cerrado regions, respectively, with 66 species occurring in both regions. However, all individuals recorded across plots (i.e., including palms and poorly sampled species) were considered in estimates of species mean relative local abundances (see below).

To accurately estimate species occupancy patterns, we additionally compiled georeferenced occurrence data for each selected plant species from the plot dataset (see above) by using the Botanical Information and Ecology Network - ‘BIEN’ R package (Maitner et al., 2017). Valid records were checked using the ‘CoordinateCleaner’ R package (Zizka et al., 2019). Synonyms and minor misspellings on specific epithets were double-checked using the ‘flora’ R package (Carvalho 2016), which follows the Brazilian Flora 2020 nomenclature (Brazil Flora Group 2019).

After the data compilation, cleaning, spatializing, and filtering steps, we calculated the local abundance and regional occupancy of each selected species, at each grid-cell scale in each biogeographic region. Local abundance was calculated as the mean relative abundance of a species across grid-cells at which it was present(hereafter, local abundance), whereas occupancy is defined as the proportion of grid-cells occupied by any given species by collapsing plot-based data and individual occurrences.

We overlaid the spatial grid-cells on global maps of key climate, soil, topographic and land-use variables. Environmental variables were then averaged within grid-cells to measure species’ niche attributes (see below). For peripheral grid-cells straddling the boundaries between biogeographic regions, averaged environmental variables only considered the area representing each region.

To describe species niches, two key measures were adopted: niche marginality and niche breadth, based on the outlying mean index (OMI) analysis (Dolédec et al., 2000), calculated using the niche function, both in the ‘ade4’ R package (Chessel et al., 2012).

Usage Notes

The dataset includes (1) woody plant species abundance data from sample plots aggregated by grid-cells of different spatial resolution, (2) occurrence data for selected species, (3) spatial coordinates of the centroids of studied grid-cells, (4) environmental variables averaged by grid-cells, and (5) summary of local abundance, occupancy, functional traits and niche attributes of selected species. 


Conselho Nacional de Desenvolvimento Científico e Tecnológico, Award: 403770/2012-2

Newton Fund

Fundação de Amparo à Ciência e Tecnologia do Estado de Pernambuco, Award: BFP-0164-2.05/19

Fundação de Amparo à Pesquisa do Estado de São Paulo, Award: 2013/08722-5