COMMENTARY Charting uncertainty about ant origins Ross H. Crozier* School of Marine and Tropical Biology, James Cook University, Townsville, Queensland 4811, Australia ver a wide range of environ- ments, up to five ant species forage every square meter of ground (1). In Amazonian Orainforests, the biomass of ants dwarfs that of vertebrates (2), and in many rainforest trees, ants make up a large fraction of individual insects (3). This ecological dominance and the complex- ity of their societies makes their phylo- geny of great interest as a glimpse into the development of the modern world in terms of the relationships between the various groups of ants, how their charac- Fig. 1. Certainty becomes managed uncertainty. Analyzing the complete sequence ant data set together teristics evolved, and when they origi- with outgroups yielded tree A, with the outgroups joining at the filled star, implying that the Leptanillinae nated. This year we have seen not one are the sister group to all other ants. Analyzing just the ant sequences led to a significantly different result, tree B. Testing nine hypotheses (dots or stars) for rooting the ant tree eliminated four but left five as but two blockbuster articles examining statistically not separable. The most likely of these, shown with a filled star, remains on the branch to the ant phylogeny and time of origin of the Leptanillinae, but the second most likely, shown with an open star, falls on the branch to the rest of the group, one of which is by Brady et al. ants, implying that the leptanillines are closely related to the Amblyponinae, with which they share some (4) in this issue of PNAS. The two arti- striking characteristics. Thick lines denote branches with posterior probability of at least 0.95. The cles (4, 5) agree in several important instability of the ingroup according to whether outgroups are included in the analysis may have resulted respects but disagree in others. from long branch attraction. Early thought on ant phylogeny was bedeviled by the belief that all or most of the genera with armored cuticles and species from all 20 currently recognized geny inference, can also lead to spurious strong stings belonged in a single sub- ant subfamilies and 10 outgroups and rearrangement of the ingroup taxa (13). family, the Ponerinae (6). Brown (7) using 6 kb of DNA sequence from seven The problem is mainly one for parsi- pointed the way forward by suggesting nuclear genes. There is much agreement mony and will not occur for maximum that various other ant subfamilies arose between the two studies. In particular, likelihood or Bayesian analysis when the within the ponerines, which are thus most subfamilies are monophyletic, and substitution model has been correctly paraphyletic; presciently, he proposed a the two trees place them in similar posi- specified (14), but the models now avail- close relationship between the Ectatom- tions. Brown’s suggestion of a strong able may not reflect reality sufficiently minae (then a ponerine tribe) and the relationship between the Ectatomminae well to avoid it in some small, but un- giant subfamily Myrmicinae [Ͼ4,500 and the Myrmicinae is not contradicted known, number of cases (15). To para- species (8)]. However, he made no no- statistically by the new findings. phrase Li (16), substitution models are menclatural change, and subsequent au- There is thus now the emergence of the naturally artificial despite the attempt to thors tended to treat the ponerines as a promise of stability in ant phylogeny, with be artificially natural. Brady et al. (4) single group. This tendency to agglom- these studies having very similar trees, but surmised that long branch attraction erate seriously compromised the ability this result includes a puzzling anomaly, might have affected the placement of to make sense of ant phylogeny, and for namely the placement of the Leptanillinae the ant groups and thus repeated the decades the procession of phylogenetic as the sister group to all other ants. Those analysis with the outgroups omitted. Sig- schemes was notable in its diversity leptanillines that have been studied are nificant differences appeared between rather than its stability. The crucial tiny, eyeless subterranean ants with an the two analyses (Fig. 1). In particular, breakthrough came from Bolton (24), army-ant lifestyle, preying on geophilo- the poneroids, a group of morphologi- who erected a host of new subfamilies morph centipedes like wolves on elk (10). cally similar subfamilies, which had and subdivided the original subfamily Their bizarre habit of the queen feeding formed a monophyletic assemblage in Ponerinae into six; although he still on hemolymph from her larvae also oc- the rooted tree, no longer did so when placed all of these together, this recog- curs in the Amblyoponinae (11), and this the outgroups were omitted. Next, nition of difference liberated phyloge- and morphological similarities raised sus- Brady et al. tested nine hypotheses for neticists to make new findings (4, 5, 9). picions that these groups are closely re- the rooting of the ant tree by constrain- Bolton’s (24) reorganization of ant lated. Having the Leptanillinae placed at ing each such link in turn and found systematics joined with the increasing the base of the tree of all ants (4, 5) is that only four of these were eliminated ease of obtaining DNA sequence, a therefore very odd. For one thing, eye- statistically. The most likely one of the moderately good fossil record, and the balling the resulting tree gives the impres- remaining five still placed the Leptanilli- rise of phylogenetic methods able to sion that the ancestral ant was eyeless and nae as the sister group to the rest of handle large data sets and estimate di- lived underground, so that the great ma- vergence dates. The first major and con- jority of ants today must have secondarily vincing effort to elucidate ant phylogeny regained eyes and moved to hunt in the Author contributions: R.H.C. wrote the paper. at a grand scale and set it in temporal open air. The author declares no conflict of interest. context was that of Moreau et al. (5) Long branch attraction (12), in which See companion article on page 18172. earlier this year. The study by Brady et groups at the ends of long branches are *E-mail: [email protected]. al. (4) is even larger, dealing with 162 wrongly placed together during phylo- © 2006 by The National Academy of Sciences of the USA www.pnas.org͞cgi͞doi͞10.1073͞pnas.0608880103 PNAS ͉ November 28, 2006 ͉ vol. 103 ͉ no. 48 ͉ 18029–18030 Downloaded by guest on September 25, 2021 antdom, but the second most likely of surface, it also opened the night; many gent of these dates and their confidence them nests them near the Amblyoponi- ants forage at night, but exceedingly few limits yields the range 105.6–143.2 Mya. nae and allows the interpretation of the flying social insects manage this (22). The two ranges overlap, but whereas that ancestral ant as an above-ground forager It is in dating the origin of ants that the of the earlier study overlaps the Jurassic, with eyes. Thus, an ant tree in accor- emphases of this year’s blockbuster ant that of Brady et al. (4) does not. However, dance with morphological expectation is articles differ most sharply. Moreau et al. Brady et al. regard the older ages as prob- among those fitting the molecular data, (5) explored a number of dating tech- lematic and hence stress a mid-Cretaceous but it is not a done deal. niques, but both Moreau et al. and Brady age for the most recent common ancestor Given the thoroughness of the phylo- et al. (4) settled on the method of penal- for ants. Even so, Brady et al. concede genetic analysis, it is striking that this ized likelihood (23) and used the same set that there may have been ants on Earth sophistication did not extend to compar- of ant fossil ages. Whereas Moreau et al. even earlier, in the form of the apparent ative analysis, for which no details have appear to have used only the minimum ant ancestors, the enigmatic fossil-only been given. Powerful Bayesian methods ages of ant fossils in their analysis, Brady Sphecomyrminae. Where this uncertainty are now available, which, unlike earlier et al. also assigned two different fixed ages leaves the association with the angio- approaches, take uncertainty in the phy- (145 and 185 Mya) to the most basal sperms is a little unclear; it seems less logeny into account (17, 18). Given the node, basically marking the origin of the problematic to have ants diversifying in wealth of morphological, ecological, so- the presence of rich angiosperm forests cial, and behavioral characteristics ants than before these arose, but homopterans present, the data now available promise There is now the would have provided the sugary secretions a further revolution in understanding all consumed by adults in preangiosperm features of ant biology. emergence of the times. But when did ants arise? Most mod- A rich interplay between systematics, ern ants (and many predatory wasps) promise of stability morphology, and molecular phylogeny have adults subsisting mostly on floral can be traced. As noted above, the sys- nectar or hemipteran exudates while in ant phylogeny. tematic decision (24), based on mor- hunting prey (or carrion) for the young. phology, to disaggregate the various These characteristics speak for an asso- groups then classified as tribes within ciation with angiosperms, which has aculeate Hymenoptera. In each article, a the Ponerinae freed molecular phyloge- been suggested as important to the ori- maximum and minimum age was esti- neticists trying to relate the subfamilies, gin of ants (19).