Cambrian Evolutionary Rates

Phenotypic and stratigraphic data were analyzed using tip-dated Bayesian approaches (20, 21) that coestimate topologies, divergence dates, and evolutionary rates. To provide multiple independent estimates of evolutionary tempo across the Cambrian, rates of evolution of discrete, meristic, and continuous data were each estimated separately across time, using unlinked epoch clock models, which assume rates vary across time slices (but are shared across all lineages in the same time slice). Thus, rates of evolution for the 107 discrete characters were estimated for the early (Series 2, 521–509 Ma), middle (Miaolingian, 509–497 Ma), and late (Furongian, 497–485 Ma) Cambrian, and likewise (separately) for the two meristic and for the six continuous traits. Alternative models of evolutionary tempo were also evaluated using Bayes factors: a strict clock (which assumes rates are constant across time slices and across lineages) and an uncorrelated relaxed clock (which assumes rates vary across lineages but not necessarily systematically across time). Parsimony analyses were also performed to test the sensitivity of the tree topologies to analytical methods and to facilitate comparison of phylogeny inferred from phenotypic characters alone and those also incorporating temporal data. Phenotypic and stratigraphic data, details of all analyses, and executable scripts are in the SI Appendix and Dryad Digital Repository (doi.org/10.5061/dryad.v7q827k).

All analyses reveal that rates of morphological evolution were homogeneous throughout the Cambrian. In the epoch clock model, rates are marginally but insignificantly higher during the early Cambrian compared with the middle and late Cambrian (Fig. 1). Accordingly, a strict (or single-epoch) clock (SI Appendix, Fig. S2)—which assumes rates are homogeneous across the entire time period spanned by the sampled fossils—fits the data better than does the epoch clock model. The relaxed clock also returned very homogeneous rates of evolution across time (Figs. 2 and 3). This time-constant rate pattern is consistent across the discrete, meristic, and continuous characters: for all three trait types, rates of evolution are very uniform across the entire Cambrian trilobite fossil record. These results parallel the finding that speciation rates among certain Cambrian Series 2 trilobites were not unusually high, although slightly elevated relative to later times (22); however, there appears to be no obvious correlation between morphological evolutionary rates and levels of intraspecific morphological variation for Cambrian trilobites (23). Given the large number of taxa sampled (relative to the number of variable characters), there is substantial phylogenetic uncertainty, many nodes have weak support, and the consensus trees differ in certain clades between analyses (Figs. 1 and 2 and SI Appendix, Figs. S2 and S3). However, the above inferences of evolutionary rates accommodate this phylogenetic uncertainty by integrating all parameter estimates and error intervals across the full pool of sampled trees.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Dated time tree of Cambrian trilobites inferred from tip-dated Bayesian analyses of discrete, meristic, and continuous traits under a multiepoch clock, which allows rates of evolution to vary across time slices. Evolutionary rates for discrete, meristic, and continuous traits were very constant across the early, middle, and late Cambrian. Notably, all three datasets failed to exhibit sharply elevated rates in the earliest time slice. Rate units are from raw BEAST (21) output; see SI Appendix, Table S1 for absolute and scaled rates. Full species names are presented in SI Appendix.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Dated time tree of Cambrian trilobites inferred from tip-dated Bayesian analyses of discrete, meristic, and continuous traits under an uncorrelated lognormal (UCLN) relaxed clock, which allows rates of evolution to vary across all individual branches. Full species names are presented in the SI Appendix.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Evolutionary rates for discrete, meristic, and continuous traits under an UCLN relaxed clock, showing that they were very constant across the early, middle, and late Cambrian. Rates have been rescaled so that the maximum rate is 1, to make the vertical axis comparable across discrete, meristic, and continuous characters.

The basal divergence in Trilobita is estimated by our dated trees to be within the Terreneuvian (Figs. 1 and 2). In our epoch clock model (Fig. 1), this divergence is consistently within the Fortunian, with the upper (older) 95% highest posterior density (HPD) interval being 541.3 Ma. This inferred origin of trilobites sometime after the Ediacaran–Cambrian boundary represents a very conservative maximum age for the group. Because this analysis did not deliberately impose any node age constraints, the rates and divergence dates for the basal portion of the tree preceding the oldest species analyzed (519 Ma) are necessarily extrapolated from estimates derived from the subsequent (preserved) trilobite fossil record. If evolutionary rates were faster before the first trilobites appear as body fossils, then the inferred rates before 519 Ma in our trees will be underestimates, and the inferred dates would be overestimates. Faster evolutionary rates would allow for the initial phenotypic disparity of trilobite fossils to be established in less time. Imposing a root node age constraint of 522 Ma (13) predictably increased rates at the base of the tree preceding the oldest fossils analyzed (519 Ma), but rates in the early, middle, and late Cambrian remained very similar (SI Appendix, Table S1).

Duration of the Cambrian Explosion

Unexpectedly homogeneous rates of morphological evolution throughout the entire Cambrian trilobite fossil record support the idea that the explosion represents a truly brief evolutionary burst that began in the Terreneuvian (at the latest) and had largely concluded by Series 2 (6, 8⇓–10, 24). Regardless of the potential biological and analytical factors responsible for fast initial rates (1, 7, 8, 25), our results provide compelling quantitative evidence that this burst had ended by 519 Ma. This time constraint is also exemplified by the well-established diversity of eumetazoans in the Chengjiang biota of China (26), which has a maximum age of 518.03 ± 0.69/0.71 Ma (27). In fact, from Cambrian Series 2 onward, taxonomic conservatism is apparent among shelly and soft-bodied faunas, further suggesting unremarkable evolutionary rates during this interval; for example, the many shared families and genera across Series 2 and Miaolingian Konservat-Lagerstätten (1, 26). Although the homogeneous rates across the Cambrian are here interpreted to indicate a rapid attainment of postexplosion normality, there is an alternative interpretation: that rates were elevated across most of the Cambrian. However, the longevity of trilobite morphotypes (e.g., genera and families) across the Cambrian (28) and general stability of faunas discussed above make the alternative interpretation less likely. Thus, Chengjiang and younger Cambrian Burgess Shale-type (BST) deposits should not be considered snapshots of the unfolding explosion but rather the early (postexplosion) records of modern-style marine ecosystems.

Despite ongoing debate over the true origins of animal phyla, our data, as well as the Ediacaran–Cambrian geochemical, body, and trace fossil records (1, 3, 9), indicate that a modern-style marine biosphere was fully established by Series 2, followed by broad-scale evolutionary stasis throughout the remainder of the Cambrian. Given the apparent paucity of unequivocal eumetazoan representatives in the Ediacaran (4, 29), it seems that many stem- and crown-group members of most bilaterian phyla had definitively appeared and diversified in ∼20 My (Fig. 4) (8, 9, 29, 30). Among these novel body plans is rampant convergence in various forms of biomineralization (24, 30) and other anatomical innovations that allowed animals increased mobility and ways of sensing their environment (1, 8). Notwithstanding the patchy Terreneuvian fossil record, it is clear that a new style of ecological network—including greatly expanded food webs and associated nutrient cycling, plus complex tiering above and below the substrate that helped reengineer the marine ecosystem—rapidly emerged during this interval (1).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Key records of early animal evolution, seawater chemistry, and exceptional fossil preservation during the latest Ediacaran to Cambrian. Temporal range of BST deposits (5, 19) shows important examples in ascending stratigraphic order: 1, Khatyspyt Formation (Siberia); 2, Chengjiang (China) and Sirius Passet (North Greenland); 3, Guanshan (China); 4, Burgess Shale (Canada) and Kaili Formation (China); 5, Wheeler Formation (United States); 6, Marjum Formation (United States). BST deposits are less common by the Furongian. Other data sources are as follows: molecular clock date for origin of Euarthropoda represents the posterior mean node age (figure 6 in ref. 31); date for origin of trilobites is the mean node age of the epoch clock model (Fig. 1); first appearance datum (FAD) for trilobite-like traces (5), rhynchonelliform brachiopods (24), echinoderms (33), and trilobite body fossils (13); seawater chemistry (30, 34); and diversity of phyla (3), classes (3), and genera (10).

A Cambrian Origin for Trilobites

A Terreneuvian origin for trilobites contradicts previous inferences of a protracted Precambrian history (2, 14⇓–16). Furthermore, our results are supported by evidence from the euarthropod trace fossil record and molecular clocks (Fig. 4). The oldest trilobite-like traces (e.g., Rusophycus) are early Fortunian in age (5, 9), and recent molecular clocks (e.g., ref. 31) place the origin of euarthropods in the late Ediacaran or earliest Cambrian. A conservative late Ediacaran root age for euarthropods still permits a Cambrian origin for trilobites, given their derived phylogenetic position within Euarthropoda (32).

The absence of Terreneuvian trilobite body fossils can be explained under two potential scenarios. The first scenario is that the fossil record is a reasonably accurate representation of early trilobite evolution, implying that rates of morphological evolution before 519 Ma were substantially faster than subsequently (SI Appendix, Table S1). This hypothesis forces rapid dispersal between widely separated paleocontinents and is difficult to reconcile with the provincialism observed in the earliest trilobites. However, if correct, this may also explain the perceived diachronism of the first trilobite fossils on different paleocontinents (12, 13), with the group potentially originating and rapidly radiating out from Siberia, West Laurentia, or West Gondwana (15, 16). The second scenario—more consistent with our results, plus trace fossil, molecular clock, and biogeographic data (5, 11)—is that the earliest trilobites (pre-521 Ma) have not been preserved or yet discovered in Terreneuvian rocks. The diversity of other skeletonized animals from a range of environments throughout the Terreneuvian (24, 30) indicates an adequate shelly fossil record. Thus, the absence of trilobites and indeed other euarthropod body fossils in Terreneuvian rocks could be explained by their nonbiomineralized exoskeletons and the unusual dearth of soft-tissue preservation (especially BST deposits) for this time interval (Fig. 4) (4, 5, 19).

The existence of nonbiomineralized trilobites in the Terreneuvian would have required multiple lineages to simultaneously converge upon a calcite exoskeleton at around 521 Ma, unless initial evolutionary rates were much faster, thus bringing their origin and fewer lineages closer to the lower boundary of Series 2 (in support of the first scenario discussed above). Synchronous biomineralization across two or more trilobite lineages is consistent with the observation that other disparate bilaterians, such as echinoderms and rhynchonelliform brachiopods, also acquired calcitic skeletons around this time (Fig. 4) (24, 33). Notably, this time coincides with a change in ocean chemistry, particularly the onset of a calcite sea (30, 34); ambient seawater chemistry influences the type of biomineral secreted at the time skeletons evolved de novo in a clade (34). These repeated patterns suggest that compelling environmental [e.g., changing Mg/Ca ratios and oxygen levels (1, 30, 35)] and/or biological factors [e.g., predation (30, 36)] were influencing this major episode of biomineralization and diversification during the final stages of the Cambrian explosion. The shelly fossil record of animals thus dramatically improves only around 521 Ma, but by that stage, the Cambrian explosion was largely over.