Sampling

We sampled all plant life forms in eight plots of 20-m x 50-m (1,000 m2) subdivided into subplots of 10-m x10-m (Figures S1, and S2). Four plots were allocated in different situations of Caatinga near road at distances from 30-m to 170-m from BR-235, and four plots in Caatinga far from roads at distances from 4,500-m to 8,500-m from BR-235, thus totaling eight plots and 0.8 ha (Figure S3). The distances were Log10 transformed for statistical analysis that used distances as continuous variable. We divided species into two components: woody (arboreal-shrubby) and non-woody (herbaceous-subshrub). All woody individuals with a circumference at ground level ≥ 10 cm and height ≥ 1 m, including woody lianas, were considered as the woody component of vegetation. The survey of non-woody species was carried out in two 10-m x 10-m subplots within each 20-m x 50-m plot (Figure S2), these being the first and last sampled subplots of the woody component (Figure S1). The cover of all non-woody species was estimated using the Braun-Blanquet method. The first five values (5 - 75 to 100% coverage, 4 -50 to 75%, 3 - 25 to 50%, 2 - 5 to 25% and 1 - 1 to 5%) of the Braun-Blanquet method refer to coverage while the last two value scales (+ - species with many individuals and low coverage, and r - rare species with low coverage) are primarily estimated by abundance (Munhoz & Araújo, 2011). From the Braun-Blanquet scale, we calculated coverage area (CA) and relative coverage (RC) using averaged values for each category (Kent, 2012).
We classified all species according to their life form using Raunkiaer’s life form system (Martins & Batalha, 2011). Species were also classified according to functional characteristics related to resilience/resistance to disturbances including resprouting ability, urticancy/toxicity and spinescent succulents, and according to functional characteristics affected by goat herbivory, including endozoochory, because of the fleshy fruits, and nitrogen fixation, because of the low C/N (Carrión et al. , 2017; Leal et al. , 2003; Severson & DeBano, 1991). Because of its dubious life form, the hemiepiphyte Hylocereus setaceus (Salm-Dyck) Ralf Bauer was considered in both, the woody and the non-woody community.
All of the species found with regrowth after a loss of aerial parts were considered resprouters, while all species with succulent and spiny structures were considered succulent spinescent (Carrión et al. , 2017). Urticancy/toxicity was considered if species exhibited some vegetative organ with urticancy or that produce some toxic compound that affects goats, sheep and cattle (Bezerra et al. , 2012). We considered nitrogen fixers as those species and genera that have a record of positive nodulation by nitrogen-fixing bacteria in the literature (Sprent, 2009). Species were considered endozoochoric through records in the literature or morphological observation (Pijl, 1982). In this study, secondary dispersion records were not considered.
In addition, we used maximum diameter and maximum height per species for the species of the woody component as quantitative functional traits. For this we calculated the mean for the three highest individuals and the three largest diameters of each species. We also used relative coverage (RC, in m2), calculated from Braun-Blanquet categories, of each non-woody species as a functional trait.
Species were recorded and later identified by expert consultation, herbaria comparison and/or specialized literature. The taxonomic system adopted was APG IV (The Angiosperm Phylogeny Group, 2016). We used information available on the Tropicos platform (Missouri Botanical Garden, 2018) and Taxonomic Name Resolution Service tool version 4.0 (Boyle et al. , 2013) for appropriate species nomenclature and respective abbreviations of the authors.