Abstract: Species flocks (SFs) fascinate evolutionary biologists who wonder whether such striking diversification can be driven by normal evolutionary processes. Multiple definitions of SFs have hindered the study of their origins. Previous studies identified a monophyletic taxon as a SF if it displays high speciosity in an area in which it is endemic (criterion 1), high ecological diversity among species (criterion 2), and if it dominates the habitat in terms of biomass (criterion 3); we used these criteria in our analyses. Our starting hypothesis is that normal evolutionary processes may provide a sufficient explanation for most SFs. We thus clearly separate each criterion and identify which biological (intrinsic) and environmental (extrinsic) traits are most favourable to their realization. The first part focuses on evolutionary processes. We highlight that some popular putative causes of SFs, such as key innovations or ecological speciation, are neither necessary nor sufficient to fulfill some or all of the three criteria. Initial differentiation mechanisms are diverse and difficult to identify a posteriori because a primary differentiation of one type (genetic, ecological or geographical) often promotes other types of differentiation. Furthermore, the criteria are not independent: positive feedbacks between speciosity and ecological diversity among species are expected whatever the initial cause of differentiation, and ecological diversity should enhance habitat dominance at the clade level. We then identify intrinsic and extrinsic factors that favour each criterion. Low dispersal emerges as a convincing driver of speciosity. Except for a genomic architecture favouring ecological speciation, for which assessment is difficult, high effective population sizes are the single intrinsic factor that directly enhances speciosity, ecological diversity and habitat dominance. No extrinsic factor appeared to enhance all criteria simultaneously but a combination of factors (insularity, fragmentation and environmental stability) may favour the three criteria, although the effect is indirect for habitat dominance. We then apply this analytical framework to Antarctic marine environments by analysing data from 18 speciose clades belonging to echinoderms (five unrelated clades), notothenioid fishes (five clades) and peracarid crustaceans (eight clades). Antarctic shelf environments and history appear favourable to endemicity and speciosity, but not to ecological specialization. Two main patterns are distinguished among taxa. (i) In echinoderms, many brooding, species?rich and endemic clades are reported, but without remarkable ecological diversity or habitat dominance. In these taxa, loss of the larval stage is probably a consequence of past Antarctic environmental factors, and brooding is suggested to be responsible for enhanced allopatric speciation (via dispersal limitation). (ii) In notothenioids and peracarids, many clades fulfill all three SF criteria. This could result from unusual features in fish and crustaceans: chromosome instability and key innovations (antifreeze proteins) in notothenioids, ecological opportunity in peracarids, and a genomic architecture favouring ecological speciation in both groups. Therefore, the data do not support our starting point that normal evolutionary factors or processes drive SFs because in these two groups uncommon intrinsic features or ecological opportunity provide the best explanation. The utility of the three?criterion SF concept is therefore questioned and guidelines are given for future studies.