Dataset preparation
A data matrix comprised by all sample sites reported in allstudies was built from the database described above. For each study captured in our database, sites where sampled populations occurred were logged in the data matrix as 0’s. Sites located between sampled populations where genetic discontinuities were recorded, and hence putative BGF existed, were logged as 1’s. Thus, for each study, whenever panmixia were detected, geographic ranges between genetically homogeneous populations were marked as a string of zeros between sample sites. From this data matrix, we built a frequency data matrix by summing all putative BGF occurrences across logged studies, normalized by the total number of studies that included each site (Table S1).
A large normalized BGF frequency value in a given site indicates that a large number of studies identified that particular site as either a BGF or a location of high probability of occurrence of BGF. Consequently, those sites can be considered locations of high BGF concordance. Conversely, a small value means either absence of BGF in a particular site (= area of high genetic connectivity) or that a small number of articles and studies included those sites within the range of their sampling designs. A distribution of the frequencies of occurrence of putative BGF along a latitudinal gradient (which is the case for most of the Brazilian coast) can be easily applied to comparative phylogeography, because this distribution represents the likelihood for which a phylogeographic break exists in a given location given by multiple empirical studies. Thus, the frequency of occurrence of putative BGF not only quantifies but also describes how one of the most important concepts in comparative phylogeography is distributed in space: concordance in the spatial distribution of phylogenetic breaks across co-occurring species (type III concordance in Avise, 2000).