Technical Comments

Comment on “Whole-genome analyses resolve early branches in the tree of life of modern birds”

See allHide authors and affiliations

Science  25 Sep 2015:
Vol. 349, Issue 6255, pp. 1460
DOI: 10.1126/science.aab1062


Jarvis et al. (Research Articles, 12 December 2014, p. 1320) presented molecular clock analyses that suggested that most modern bird orders diverged just after the mass extinction event at the Cretaceous-Paleogene boundary (about 66 million years ago). We demonstrate that this conclusion results from the use of a single inappropriate maximum bound, which effectively precludes the Cretaceous diversification overwhelmingly supported by previous molecular studies.

Substantial discrepancies exist between the first appearance in the fossil record of many avian lineages and their age as estimated by molecular clock analyses (1). A particularly contentious topic is the time scale for the divergences between modern bird orders (2). The oldest fossil generally agreed to fall within Neornithes—the clade comprising all living birds (Fig. 1A)—is Vegavis (3), from the end-Cretaceous ~67 million years ago (Ma). However, most molecular dating studies have suggested that Neornithes began diversifying more than 110 Ma (46). Conversely, Jarvis et al. (7) presented genome-based molecular clock results that supported predominantly post-Cretaceous dates for the diversification of modern birds.

Fig. 1 Reanalysis of Jarvis et al.’s genomic data set.

(A) Jarvis et al.’s TENT topology (partially collapsed for clarity) placed into temporal context with the earliest records for several older avian lineages. Divergence dates among living species are derived from analyses following Jarvis et al. (7) but employ a more appropriate 117.5-Ma maximum bound for the diversification of neornithines. In addition, ancestral habitat was reconstructed under maximum parsimony for terrestrial (orange) and aquatic (blue) lineages, with approximate transition times based on fossil records. Dotted lines indicate the existence of derived aquatic taxa within a clade for which terrestriality appears ancestral. (B) Posterior age distributions for several key nodes under three different maxima for Neornithes: 99.6 Ma (dotted line), 117.5 Ma (dashed line), and removal of this maximum constraint (solid line).

Although Jarvis et al.’s genomic data set and phylogeny are impressive leaps forward for the field, the influence of their calibrations requires closer investigation. Their phylogeny was calibrated by applying minimum and maximum bounds to key nodes. Best-practice approaches for defining minimum bounds are well established (8, 9): The origin of a clade necessarily predates the appearance of its first unequivocal fossil representative. However, maximum bounds (the earliest plausible age of a clade) are a more difficult proposition, yet just as important for estimating dates (10, 11).

Jarvis et al.’s only maximum age constraint among birds (the root of Neornithes) was weakly justified. They implemented a strong prior against the diversification of Neornithes occurring before the Late Cretaceous (99.6 Ma). This decision effectively precluded results consistent with many previous molecular clock studies, which have generally recovered ages for Neornithes between 110 and 140 Ma (46). Their first justification for the 99.6-Ma bound, that it is 30 million years older than their age estimate for Neoaves (two nodes shallower), is immaterial. Their second justification, that the bound “far exceeds the age of paleontological evidence for the existence of Neornithes” [supplementary materials for (7)] [e.g., the putative neornithine, Austinornis at ~85 Ma (12)], is only relevant if it covers relatively well-sampled fossil assemblages in potential geographic areas of origin that contain no putative clade members but do preserve ancestral forms or ecological equivalents (10, 13). However, the few fossils attributable to Euornithes (neornithines and their close relatives) (see Fig. 1A) during this period (from ~85 Ma to Jarvis et al.’s maximum at 99.6 Ma) are taphonomically biased, including only a handful of seabirds (14). Because early neornithines may well have been terrestrial (Fig. 1A), assigning a maximum constraint based on this Late Cretaceous record is clearly inappropriate.

We reanalyzed Jarvis et al.’s molecular dating data set (722,202 nucleotides from 51 diapsid species, including 48 birds) using the authors’ total evidence nucleotide tree (TENT) topology, while varying the maximum bound for Neornithes. Our analyses were otherwise performed according to the methods described in the original study.

Recent studies have used the absence of neornithines from some Early Cretaceous deposits, where the avian fossil record is far superior, to set a 97.5% prior maximum bound of 117.5 Ma (5). When we reanalyzed Jarvis et al.’s genomic data with this more appropriate maximum bound, the number of sampled neornithines with pre–Cretaceous-Paleogene (K-Pg) mean divergence estimates increased from 15 to 22 out of 47 [nodes with 95% highest posterior densities (HPDs) entirely pre–K-Pg increased from 10 to 13], and the posterior mean for the origin of modern birds fell close to the maximum, at 117.2 Ma (Fig. 1B).

However, even this older maximum constraint remains inappropriate if modern birds were initially restricted to Gondwana, as is presently the case for most basal neornithine clades, including ratites, megapodes, cracids, and screamers. Within Late Cretaceous Gondwana, the oldest bird fossil is a putative neornithine (15), and the known Early Cretaceous Gondwanan fossil record is largely uninformative, containing only the non-euornithine Nanantius and taxonomically indeterminate fragments. When such potential geographic origins are considered, it becomes difficult to justify even a 117.5-Ma maximum. Removing the maximum bound on Neornithes altogether resulted in pre–K-Pg mean estimates for a full 37 of our 47 neornithine divergences (29 nodes with 95% HPDs entirely pre–K-Pg), and the posterior mean for the origin of modern birds increased to 162.5 Ma (Fig. 1B). This date is substantially older even than Archaeopteryx, suggesting that substitution or rate variation models are misspecified.

In summary, Jarvis et al. (7) employed a strong maximum constraint on the age of Neornithes that is poorly supported by the fossil record and that heavily influenced their inference of post-Cretaceous bird diversification. Without imposing this constraint, their genomic data and remaining calibrations favor deep Neornithes origins, with most interordinal divergences occurring in the Cretaceous, consistent with previous molecular dating studies. If indeed the major diversification of modern birds occurred after the K-Pg mass extinction event, which arguably better fits the fossil record, it could be explained by parallel rate decelerations across modern avian clades being nonidentifiable even to relaxed clock methods. However, support for such agreement with fossil records must come from improvements in either the model of molecular rate variation or better fossil sampling and not from the imposition of artificially strong maximum age constraints.

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article