INTRODUCTION
The Paleocene-Eocene Thermal Maximum (PETM) [~56 million years (Ma) ago] was an episode of global warmth brought about by the rapid [<5 thousand years (ka) (
1)] release of up to 10,000 gigatons (
2) of carbon into the atmosphere from a sedimentary (
3) and/or volcanic (
2) source. Global marine temperatures rose by 5° to 8°C (
4), and acidification in the mixed layer (
2,
5) and deep ocean (
6) was severe and often coupled with hypoxia both at depth (
7) and on continental margins (
8). The PETM likely underestimates the expected impact of ongoing combustion of fossil fuels (
2,
9) but nonetheless remains the closest analog to the present offered by the geological record. As such, the interval has received intense scrutiny for insights on how the present-day climate system and biosphere might react to projected anthropogenic increases in atmospheric CO
2. The evolutionary, ecological, and biogeographic responses of terrestrial faunas and floras, planktonic ecosystems, and the deep-sea benthos have all been documented in some detail (
10), yet very little is known about the impact of the PETM on marine shelf macrofaunal assemblages (
11). The dearth of known macrofossiliferous shallow-marine sections, compounded by the low probability of preserving such a short-duration event in complex depositional settings rife with small-scale hiatuses (
12), continues to hinder interpretations.
The U.S. Gulf Coastal Plain (GCP) contains a thick, highly fossiliferous, and much-studied section spanning the Paleogene and, as such, provides one of the best overall records of marine mollusk assemblage change in shallow-shelf environments through this interval in the world (
13). The PETM has been located in the section based on dinoflagellate biostratigraphy and a diagnostic carbon isotope excursion (CIE) (
14), and while marine macrofossils are not known from the CIE—and hence the interval of maximum warmth—itself, rich shell beds closely bracket the interval. Despite an inability to see the immediate short-term response of mollusks to PETM CO
2 release, a comprehensive comparison of faunas before and after can reveal the long-term evolutionary impacts of this transient environmental perturbation. In the same way that other biotic perturbations leave a lasting impact on the history of a group, as evidenced by delayed recovery of richness, peaks in first and last appearances, lasting shifts in ecologic structure and composition, or persistent changes in body size, we should expect to see the fingerprint of the PETM on the evolutionary and ecological history of mollusks, even while an assessment of its immediate impact remains elusive. We therefore bring a range of tools to bear on the rich molluscan fossil record of the GCP to explore whether the PETM had a significant lasting effect on the richness, turnover, or ecological structure of shelf faunas or the body size, growth rate, or life span of component taxa.
DISCUSSION
Taken as a whole, our results indicate that the long-term impact of the PETM on these shallow-water benthic communities was unremarkable. Unlike the deep-sea benthos (
22), molluscan shelf associations on the GCP suffered little in the way of lasting biodiversity loss, taxonomic turnover, or persistent ecological restructuring relative to changes at earlier or later formation boundaries. Turritellines show a shift toward smaller body size, a prediction associated with both warming (
24) and acidification (
25), but the magnitude of change seen in this group is not matched by other important groups. Growth rates of the most common turritelline and venericardiine species do show pronounced change across the PETM, but in opposite directions. This lack of substantial or generalizable pattern in faunas that straddle the PETM suggests that any potential selection pressure imparted by environmental changes during the PETM must have been weak, taxon-specific, and/or short-lived, and ultimately inconsequential to overall molluscan evolutionary history.
One potentially selective pattern present in our data set is the statistically significant shift to infauna-dominated assemblages after the PETM, a trend documented in proportional abundance (
Fig. 3 and fig. S2), but not in richness (fig. S3). This change could have been driven by selection during the PETM such that earliest Eocene faunas exhibit the relict consequences of that ecological reorganization. During the PETM, sediments of the U.S. Atlantic Coastal Plain (ACP) (
8,
26) and GCP (
14) both suggest at least seasonal hypoxia and eutrophication along continental margins, an increase in nutrients delivered with runoff, and remobilization of phosphate in sediments. Epifaunal organisms generally tend to be more sensitive to hypoxia than infauna (
27), and infaunal bivalves are better suited to conditions of increased primary production (
28), consistent with observed patterns. A trend toward increasing richness and abundance of infaunal suspension feeders and decreasing richness of epifaunal suspension feeders characterizes the Cretaceous and Paleogene section in the GCP overall (
17,
20), but a marked drop in the abundance (but not richness) of epifauna coupled with an increase in abundance (but not richness) of infauna suggests a departure from the longer-term evolutionary trends, perhaps reflecting a more immediate response to PETM ecological forcing via hypoxia that is carried over into earliest Eocene assemblages. Turritellines dominate the semi-infaunal guild, and living species are known to decrease in biomass and abundance in association with hypoxia (
29), consistent with their observed lower abundance and smaller body size in post-PETM assemblages. We are likewise tempted to ascribe significance to the proportional increase in both richness and abundance of chemosymbiotic bivalves following the PETM, as a chemosymbiotic ecology should be favored in hypoxic sediments with more active sulfate reduction (
30). Kosnik (
20) cautions that there is a good deal of heterogeneity in both the richness and abundance of chemosymbiotic taxa among formations in the Cretaceous-Paleogene GCP. Despite this, the appearance in our data set of seven species in five different genera from two families, together comprising a small but significant percentage of individuals [4.3 ± 0.003% (1 SD)], in the earliest Eocene GCP is perhaps notable when late Paleocene units each contain only one species at lower abundance. Like the shift toward infauna, this might also be a relic of selection favoring hypoxia-tolerant taxa during the PETM itself. Note that while chemosymbiotic taxa are also infaunal, they comprise only a small proportion of occurrences (6.8%) and individuals (6.5%) within the earliest Eocene infauna, and so do not themselves control the trend toward infaunalization.
Acknowledging the potential for some lingering effects of PETM hypoxia-driven selection in the earliest Eocene faunas, by and large the mollusk faunas of the GCP offer little sign of impact from the marked but short-lived environmental changes that transpired during the PETM globally. Work on the foraminiferal assemblage of the ACP suggests a similarly muted response in the long term, with almost no taxonomic turnover on the shelf (
26). When viewed at a comparable temporal resolution (assemblages from before and after, but not during, the CIE), our mollusk record from the GCP and the foraminiferal record from the ACP shelf (
26) look very much like the long-studied plant fossil record from the western United States—each shows no major lasting change. It was only after floras (or foraminiferans on the ACP) were discovered within the CIE in sections that fortuitously captured the event that the full impact of the PETM became apparent (
31). A similarly significant yet transient biogeographic, ecophenotypic, and/or ecologic response is still possible and even probable in marine shelf mollusks as well, but the stratigraphic record of the GCP as yet does not allow us to test this hypothesis.
The absence of a notable P-E (Paleocene-Eocene) boundary peak in first or last appearances of marine macrofauna in a global Phanerozoic database (
32) suggests that the GCP pattern is not atypical of marine shelf assemblages more broadly. If so, why is it that shelf macrofauna were evidently less affected in the long term by PETM events? Several factors could be involved, including historical “preconditioning” of faunas to warm temperatures and the comparatively slow rate of CO
2 release that limited the acidification of surface waters. With respect to the former, organisms at Earth’s surface were already adapted to the warm conditions of the Paleocene, and with low equator-to-pole thermal gradients, a good deal of the planet likely experienced occasional summer temperatures that approached PETM-like warmth. Benthic mollusks, with multiyear life spans and essentially no ability to migrate, would have had to endure these episodes, fortuitously preparing them to emerge unscathed from the PETM itself. The only other shelf macrofaunas explicitly studied across the PETM are those from the tropics, where sea surface temperatures may have exceeded 36°C (
33)—reef corals were replaced by larger benthic foraminiferans (
11) and reef fishes experienced a major evolutionary turnover (
34), having a lasting impact on the history of both groups. While tropical planktonic taxa, with the capacity for rapid biogeographic range shifts, were able to recover rapidly despite short-term precipitous drops in abundance and diversity (
33), at least some macrofauna evidently suffered more lasting effects. Whether tropical mollusks experienced a similar perturbation is unknown, but our data show that GCP subtropical assemblages, at least, escaped relatively intact.
In addition to warming, ocean acidification is an anticipated consequence of the addition of CO
2 to the atmosphere (
35), and one with potentially serious implications for marine ecosystems (
36). Carbonate dissolution is a prominent feature of deep-ocean PETM sediments (
6), but carbon cycle modeling suggests that CO
2 release was sufficiently slow that transfer to the large deep-sea reservoir via overturning circulation limited buildup in the atmosphere and surface ocean (
2). Recent work has provided proxy evidence for PETM sea surface acidification [0.3 to 0.4 pH units or approximately 100% more acidic (
2,
5)], but a biological response has yet to be demonstrated, even in the most sensitive of carbonate skeletal elements (
8,
37). PETM mollusks would have been most vulnerable as larvae (
36), but a comprehensive metadata analysis (
36) of modern mollusks found minimal to no impact on juveniles or adults experiencing a drop of <0.4 pH units, at least on a time scale of years or less. Hence, the PETM drop in mixed-layer pH was evidently insufficient to hinder calcification or leave a lasting imprint on patterns of survivorship or body size in these skeletal invertebrates. Buffering of coastal waters by more alkaline runoff sourced from enhanced continental weathering in the aftermath of the PETM may have additionally helped to spare shelf ecosystems (
38), but this process would not be instantaneous even on shelves, providing thousands of years for selection to be effective had forcing been important.
What, if any, implications do these results hold for the present and future response of shallow marine biota to ongoing global change? The rate of anthropogenic CO
2 release to the atmosphere today is an order of magnitude greater than that during the PETM (
2). Given the amount of carbon released to date and even the conservative projections for the future, the degree of warming, acidification, and hypoxia of the surface ocean associated with the PETM is likely to be a “best-case” scenario for what to expect in our geologically near future (
9). The markedly faster rate of environmental change today combined with the fact that modern organisms are instead adapted to a cooler climate highlight the differences between the PETM and the present day. The minimal response on long (geological) time scales to a perturbation that took place over several thousands of years does not imply that modern shelf ecosystems are not at risk from the climate change playing out over the next few hundred years.
Acknowledgments
We thank D. Dockery from the Mississippi Department of Environmental Quality for providing and helping with additional bulk sample material, for assisting in the field on multiple occasions, and for his long-standing support of research on GCP paleontology. A number of students helped with sample processing, size measurements, and data entry, including E. Judd, M. Kosloski, U. Smith, P. Stoddard, L. Eccles, E. Fenlon, R. Lee, I. Sobalvarro, K. Ohman, D. Reed, K. McClure, R./T. Lee, and T. Schlossnagle. B. Gollands constructed the data entry portal for assembling collection lists from the Palmer and Brann monograph. Thanks to L. Ward (Virginia Museum of Natural History) for help in the field. E. Thomas, C. Jaramillo, J. Jackson, two anonymous reviewers, and members of the PaleoX seminar group at Syracuse University read the manuscript and provided much helpful feedback. This represents PaleoDB paper #314.
Funding: Research was supported by collaborative NSF Earth Sciences awards to L.C.I. (0719645), W.D.A. (0719642), and R.L. (0718745).
Author contributions: L.C.I., W.D.A., and R.L. conceived the project, secured funding with J.A.S., conducted field work, and oversaw student contributions. J.A.S. and C.P. processed and tabulated the bulk sample data. J.A.S. and L.C.I. tabulated monographic works into occurrence-based collections and archived the data with the PaleoDB. J.A.S. and C.P. extracted and tabulated all the faunal data from the PaleoDB. C.P. performed guild analyses and sampled venericardiines for stable isotopes. R.L. oversaw the body size and phylogeny work on venericardiines. W.D.A. oversaw the body size and phylogeny work on turritellines. L.C.I. oversaw life history analyses using stable isotopes. J.C.H. wrote and implemented the R code and performed all statistical analyses, except those associated with body size. L.C.I. and C.P. wrote the manuscript and drafted the figures. All authors read the manuscript and provided edits.
Competing interests: The authors declare that they have no competing interests.
Data and materials availability: New and existing taxonomic data for these analyses reside in the PaleoDB (
paleobiodb.org); collections used are linked to this paper (PaleoDB reference #34008) and were derived largely from the works of K. W. Palmer and D. C. Brann, L. D. Toulmin, and J.A.S. All data and R codes are also provided in full in the Supplementary Material files associated with this paper or may be requested from the authors.