Research Article

Exceptional continental record of biotic recovery after the Cretaceous–Paleogene mass extinction

See allHide authors and affiliations

Science  22 Nov 2019:
Vol. 366, Issue 6468, pp. 977-983
DOI: 10.1126/science.aay2268

Terrestrial record of recovery

The extinction that occurred at the end of the Cretaceous period is best known as the end of the nonavian dinosaurs. In theory, this paved the way for the expansion of mammals as well as other taxa, including plants. However, there are very few direct records of loss and recovery of biotic diversity across this event. Lyson et al. describe a new record from the Cretaceous-Paleogene in Colorado that includes unusually complete vertebrate and plant fossils that describe this event in detail, including the recovery and expansion of mammalian body size and increasing plant and animal biotic diversity within the first million years.

Science, this issue p. 977


We report a time-calibrated stratigraphic section in Colorado that contains unusually complete fossils of mammals, reptiles, and plants and elucidates the drivers and tempo of biotic recovery during the poorly known first million years after the Cretaceous–Paleogene mass extinction (KPgE). Within ~100 thousand years (ka) post-KPgE, mammalian taxonomic richness doubled, and maximum mammalian body mass increased to near pre-KPgE levels. A threefold increase in maximum mammalian body mass and dietary niche specialization occurred at ~300 ka post-KPgE, concomitant with increased megafloral standing species richness. The appearance of additional large mammals occurred by ~700 ka post-KPgE, coincident with the first appearance of Leguminosae (the bean family). These concurrent plant and mammal originations and body-mass shifts coincide with warming intervals, suggesting that climate influenced post-KPgE biotic recovery.

The Cretaceous–Paleogene (K–Pg) boundary marks Earth’s most recent mass extinction, when >75% of species, including nonavian dinosaurs, went extinct (1). In the terrestrial realm, the mass extinction was followed by a radiation of modern clades, particularly placental mammals (2), crown birds (3), and angiosperms (4). The drivers (58) and tempo (9, 10) of the K–Pg mass extinction (KPgE) have been hotly debated, and the patterns of terrestrial recovery in the first million years after the KPgE remain poorly understood. The extinction of all large-bodied vertebrates (5) undoubtedly affected the post-KPgE taxonomic, ecologic, and body-mass diversification of various clades, but the lack of a well-preserved fossil record has left the factors influencing ecosystem recovery unknown. Here, we provide a detailed and temporally constrained terrestrial fossil record from this critical interval.

Fossils of terrestrial and freshwater organisms from the first million years after the KPgE are exceedingly rare worldwide, hindering our knowledge of post-KPgE taxonomic and ecological radiations. Thus far, the most fossiliferous sections from this time interval occur in the Williston, San Juan, Hanna, and Denver basins along the eastern margin of the Rocky Mountains in North America (11, 12). In all of these study areas, discontinuous outcrops result in composite stratigraphic sections; plant fossil localities are geographically widely spaced, vertebrate-bearing horizons are sparse and separated by long temporal gaps, complete vertebrate fossils are exceptionally rare, and age control is variable (1017). The Williston Basin has the most comprehensive fossil record with excellent age control, but the vertebrate specimens are fragmentary (10, 12, 13). The San Juan Basin preserves a well-studied early Paleocene vertebrate record but does not record the K–Pg boundary itself (16). Moreover, overlying Paleocene rocks only contain two vertebrate fossil–bearing horizons in the first 1 million years post-KPgE (16). The Hanna Basin K–Pg section is rich in fragmentary vertebrate fossils but has structurally complex strata and lacks a detailed chronostratigraphic framework (17). Finally, the Denver Basin has well-documented Cretaceous and Paleocene strata, a precisely dated K–Pg boundary, and abundant, geographically dispersed plant fossils but, prior to this study, contains only a sparse and fragmentary vertebrate fossil record (14, 15, 18, 19).

Corral Bluffs study area, Denver Basin, Colorado, USA

We developed a high-resolution stratigraphic framework in the Corral Bluffs study area (east of Colorado Springs), a single continuous (physically traceable) ~27-km2 outcrop from the Denver Basin that preserves the biotic recovery of a terrestrial ecosystem in the first million years post-KPgE (20) (Fig. 1 and fig. S1). This stratigraphy is tied to the geomagnetic polarity time scale (GPTS 2012) using paleomagnetics and chemical abrasion isotope dilution thermal ionization mass spectrometry (CA-ID-TIMS) U-Pb-dated volcanic ash (20). For comparison, ages using an alternative age model based on work in the Denver Basin (18) are also provided in data S1 to S14. The study area contains an exceptionally dense vertebrate (299 localities) and megafloral (65 localities) record, with fossils occurring at >150 stratigraphic levels in the ~250-m-thick sequence (Fig. 1). The extensive and nearly continuous outcrop belt spans the last ~100 thousand years (ka) of the Cretaceous and the first ~1 million years (Ma) of the Paleocene. It includes four North American Land Mammal Age (NALMA) interval zones, four palynostratigraphic biozones, three magnetochron boundaries, two U-Pb radiometric dates, and the palynologically defined K–Pg boundary, yielding a locally derived, high-resolution chronostratigraphic framework (Fig. 1, figs. S2 to S5, and supplementary materials) (20). Together, these data provide an unprecedented opportunity to assess the biotic recovery of a terrestrial ecosystem after the KPgE.

Fig. 1 Temporally calibrated stratigraphic, floral, and faunal data for the K–Pg interval in the Corral Bluffs study area (fig. S1).

Stratigraphy is tied to the GPTS 2012 using paleomagnetics (P-mag) and CA-ID-TIMS U-Pb-dated ash (italicized dates) (20) (data S1 and figs. S3 and S5). The composite lithostratigraphic log (figs. S2 to S5) is dominated by intercalated mudstone and sandstone, reflecting a variety of fluvial facies. Pollen zones (data S3) are defined by diversification of Momipites spp. (fossil juglandaceous pollen) (Fig. 3I). The K–Pg boundary is demarcated by the decrease in abundance of Cretaceous pollen taxa (labeled “K-taxa”) without recovery, and the subsequent fern (Cyathidites spp.) spike (data S2). Relative abundance (%) of fern (Cyathidites spp.) and palm (Arecipites spp.) (Fig. 3E) palynomorphs increased considerably post-KPgE (data S2); note that palm pollen percentages are offset from scale by 20%. Standing richness of dicot morphospecies or megafloral standing richness is exclusive of species that occur at a single locality (data S4 to S7). Leaf-estimated mean annual temperature (LMAT) calibrated with East Asian forests (data S8 and fig. S6). Horizontal pink shading indicates hypothesized warming intervals. Estimated leaf mass per unit area (data S9 and S10 and fig. S7) is shown with box plots that represent the distribution of species-site pair means for each 30-m bin (supplementary materials). Box plots are placed along the y axis near each bin’s stratigraphic midpoint and are repositioned for visibility. See data S11 and supplementary materials for placement of NALMAs. Tick marks for P-mag, pollen zones, megafloral standing richness, and NALMAs show stratigraphic placement of samples and fossil localities (supplementary materials).

Vertebrate fossils in the Corral Bluffs succession are unusually complete for this time period, are found in a range of depositional environments, and represent a diversity of taxa and body sizes (Figs. 1 and 2). Most are three-dimensionally preserved in hydroxyapatite concretions and are found in all observed facies, often as articulated skeletons or skulls with intact delicate structures such as middle ear and hyoid elements (Fig. 2). Among vertebrate specimens preserved in concretions, mammalian, turtle, and crocodilian crania (Fig. 2, A to T) and turtle shells (Fig. 2, U to X) are the most common. Individual fossils range in size from ~3 mm2 (isolated teeth) to larger forms such as 1.5-m-long, articulated crocodilian skeletons. Plant fossils also span the size spectrum across all observed facies, including microscopic palynomorphs, seeds, leaves, roots, branches, in situ saplings, and even large stumps and logs (Fig. 3).

Fig. 2 Representative selection of vertebrate fossils.

(A to L) Mammalian crania in dorsal and ventral views of Eoconodon coryphaeus [(A) and (B), DMNH.EPV.130976]; Ectoconus ditrigonus [(C) and (D), DMNH.EPV.130985]; Loxolophus sp. [(E) and (F), DMNH.EPV.132501]; juvenile E. ditrigonus [(G) and (H), DMNH.EPV.132515]; Carsioptychus coarctatus [(I) and (J), DMNH.EPV.95283]; and Taeniolabis taoensis [(K) and (L), DMNH.EPV.95284]. (M and N) Crocodilian cranium in dorsal and ventral view of cf. Navajosuchus [(M) and (N), DMNH.EPV.48541]. (O to T) Turtle crania in dorsal and ventral views of Axestemys infernalis [(O) and (P), DMNH.EPV.132514]; Palatobaena sp. [(Q) and (R), DMNH.EPV 134081]; and [(S) and (T)] Cedrobaena putorius (DMNH.EPV.130982). (U to X) Turtle shells in dorsal and ventral views of Gilmoremys sp. [(U) and (V), DMNH.EPV.95454] and Hoplochelys sp. [(W) and (X), DMNH.EPV.95453]. All crania and shells are shown to scale, except for (W) and (X), which are scaled 1:2 compared with the other specimens. Scale bar, 10 cm.

Fig. 3 Representative selection of plant fossils.

(A) In situ tree stump. (B to E) Palm fossils, including (B) in situ stump, (C) frond, (D) flower (DMNH.EPI.45594), and (E) Arecipites sp. pollen grain. (F and G) Most common smooth and toothed dicot morphospecies: (F) “Rhamnus” goldiana (DMNH.EPI.52262) and (G) Platanites marginata (DMNH.EPI.23281). (H and I) Walnut family fruit and pollen: (H) Cyclocarya sp. (DMNH.EPI.52272) and (I) Momipites tenuipolus pollen grains preserved as a dyad. (J and K) Legume: (J) seedpod (DMNH.EPI.45540) and (K) leaflet (DMNH.EPI.45576). Rock hammer handle shown in [(A) to (C)] is 38 cm long; (D) flower is 5 mm wide; (E) pollen grain is 42 μm long; (I) each pollen grain has a 20-μm diameter; leaflet in (K) is scaled 2:1 compared with (J). Scale bar, 5 cm.

We recognize 16 mammalian taxa, 8 of which are based on cranial remains, including the first occurrence of the late Puercan (Pu3) index taxon Taeniolabis taoensis (Fig. 2, K and L) from the Denver Basin. Cranial size and lower first molar area were used to estimate mammalian body mass—an important feature that affects many aspects of the biology and ecology of mammals (Fig. 4) (21). Given that there appears to be bias toward large vertebrates in our dataset (supplementary materials and data S11), we focused on maximum mammalian body mass. The largest-bodied mammals disappeared at the K–Pg boundary (10) and returned to near pre-KPgE levels within 100 ka after the K–Pg boundary (Fig. 4). Subsequent shifts in maximum mammalian body mass occurred at the Pu1–Pu2 and near the Pu2–Pu3 transitions, ~300 and ~700 ka post-KPgE, respectively (Fig. 4). In addition, the pattern and abundance of vertebrates preserved in all paleoenvironments suggest that by ~700 ka post-KPgE, the largest mammals (25+ kg) were spatially partitioned across the landscape. We observe a strong pattern of association between taxa and facies (Fig. 4), indicating that baenid turtles (Fig. 2, Q to T) and T. taoensis (Fig. 2, K and L) lived in or near river channel margins, whereas chelydroid turtles (Fig. 2, W and X) and the large periptychid mammals Ectoconus ditrigonus (Fig. 2, C, D, G, and H) and Carsioptychus coarctatus (Fig. 2, I and J) primarily occupied distal portions of the floodplain (Fig. 4).

Fig. 4 Timeline of expansion of maximum body mass and niche space in earliest Paleocene mammals correlated with diversification and origination of key plant groups and warming intervals.

Post-KPgE “disaster” ecosystems occur for <100 ka, ecosystem “recovery” occurs between ~100 and 300 ka, and overall post-KPgE ecosystem equilibrium occurs within ~300 ka. Mammalian body mass estimated on the basis of cranial and lower first molar dimensions of specimens recovered from Puercan 1 to Puercan 3 (Pu1–Pu3) intervals (data S13 and S14 and figs. S8 and S9). Data from Corral Bluffs study area (yellow) except for Pu1 mammals, which come from adjacent outcrops in the Denver Basin [West Bijou (orange), South Table Mountain (blue), and Alexander Locality (green)] and Didelphodon from North Dakota (red) (data S13 and S14 and supplementary materials). Not plotted is the distribution of other large (10 to 100+ kg) vertebrates (e.g., turtles, crocodilians, dinosaurs) found throughout the section (Fig. 1). Horizontal pink shading represents hypothesized warming intervals interpreted from LMAT. Niche partitioning graph showing environmental distribution of vertebrate groups (data S12): Carsioptychus, Ectoconus, and chelydroid turtles predominantly associated with floodplain and ponded water facies; baenid turtles and Taeniolabis predominantly in river channel complexes and proximal to medial crevasse splay facies. FAD, first appearance datum; m1, first molar tooth.

We recognize 233 plant morphospecies in our study area (supplementary materials). Despite fewer samples from Cretaceous strata (11 Cretaceous localities versus 54 Paleocene localities), richness of dicotyledonous (dicot) leaf morphospecies from raw species counts at localities in the last ~100 ka of the Cretaceous (−18 to 0 m; 7 localities, 777 specimens, most speciose locality n = 31) and the first ~100 ka of the Paleocene (0 to 20 m; 6 localities, 1019 specimens, most speciose locality n = 13) indicates that earliest Paleocene dicot diversity was less than half that of the latest Cretaceous (fig. S6). Additionally, 46% of Cretaceous dicot leaf morphospecies that occur at more than one site do not occur in any of our Paleocene localities. A comparable study with similar time bins from the Williston Basin estimated 57% extinction in dicot leaf morphospecies at the KPgE (22). Leaf mass per area (LMA), a proxy for carbon investment and ecological strategy in plants (23), decreased in both maximum and minimum values across the K–Pg boundary (Fig. 1 and fig. S7), which is consistent with a shift to faster growth strategies. Megafloral standing richness and LMA are lowest in the earliest Paleocene but exceed pre-KPgE levels within ~300 ka (Fig. 1 and fig. S7).

After the KPgE, many angiosperm clades diversified (4). The Corral Bluffs section preserves the oldest known occurrence of the Leguminosae, or bean family, represented by fossil seedpods and leaflets dated to 65.35 Ma (Fig. 3, J and K). The oldest previously recognized legume (24) is based on wood and leaflets (25) from early Paleocene rocks of Argentina (26), whereas the earliest legume seedpods are not recognized until the late Paleocene (~58 Ma) of Colombia (27). Our discovery supports (i) a nearly synchronous first appearance of legumes in North America and southern South America; (ii) a rapid diversification for the group in the earliest Paleocene (24); and (iii) their apparent origination in the Western Hemisphere.

Relative changes in leaf-estimated mean annual temperature (LMAT) (Fig. 1, fig. S6, and supplementary materials) from our section track paleotemperature proxies from sections elsewhere in the world. Corral Bluffs experienced a 4.6°C cooling [22.1° ± 2.7°C one standard error (1SE) to 17.5° ± 3.4°C 1SE] during the last ~100 ka of the Cretaceous, comparable to cooling estimates derived from LMAT (28) and carbonate-clumped isotopes (29) from the Williston Basin, and δ18O of benthic foraminifera from the South Atlantic (30). For the first time, we corroborate (31) a warm interval immediately after the K–Pg boundary in a terrestrial section. Here, we observe that a 5.1°C warming event (17.5° ± 3.4°C 1SE to 22.6° ± 3.5°C 1SE) occurred from the K–Pg boundary through the first ~60 ka of the Paleocene, similar to the ~5°C in ~100 ka warming pulse inferred from δ18O of phosphatic fish scales from the El Kef K–Pg section of Tunisia (31). A second ~150-ka interval (65.80–65.65 Ma) shows an initial warming of 2.2°C (21.1° ± 3.3°C 1SE to 23.3° ± 2.9°C 1SE) over ~30 ka, sustained temperatures for ~50 ka, and then 3.0°C cooling (22.7° ± 2.8°C 1SE to 19.7° ± 3.1°C 1SE) over ~70 ka at the top of magnetochron C29r. This event corresponds with the Danian C2 carbon isotopic excursion and inferred warming interval observed in marine (32) and terrestrial (33) strata. Sampling between these warming intervals is limited, and an alternative hypothesis is a general warming trend from the K–Pg boundary to the magnetochron C29r–C29n boundary. A third 2.9° to 3.2°C warming pulse (18.0° ± 3.3°C 1SE to 20.9° ± 3.0°C 1SE to 17.7° ± 3.5°C 1SE) over ~10 ka is tentatively recognized at ~700 ka post-KPgE.

Paleotemperature and ecosystem recovery

The timing of the early Paleocene warming intervals corresponds with changes in plant richness and taxonomic composition and, likely owing to additional food sources, coincident shifts in mammalian taxonomic composition, ecologic diversification, and expansion in the range of maximum mammalian body mass (Fig. 4). A mammalian taxonomic increase has been documented elsewhere in the Denver Basin, within the first 100 ka of the Paleocene, from 9 species found in the earliest Pu1 faunas to 21 species found in later Pu1 faunas (34, 35). Maximum mammalian body mass increased through this interval to nearly pre-KPgE levels, from the largest known Lancian mammal (~8 kg) to the largest known Pu1 mammal (~6 kg), coincident with the first post-KPgE warming episode (Fig. 4 and figs. S8 and S9). The Pu1–Pu2 transition occurred ~300 ka after the KPgE and was marked by the appearance of varied and large (20+ kg) periptychid mammals. The appearance of larger-bodied periptychid mammals, particularly the herbivorous, hard-object feeder C. coarctatus (Fig. 2, I and J) (36, 37), marks a notable dietary niche specialization in the earliest Paleocene, moving from the largely omnivorous or insectivorous diet found in Pu1 mammals (38) to a more herbivorous diet found in some Pu2 mammals. This dietary shift is correlated with a threefold increase in maximum mammalian body mass compared with Pu1 faunas (Figs. 1 and 4 and figs. S8 and S9). The Pu1–Pu2 transition was coincident with the onset of a high plateau in megafloral standing richness, an increase of LMA beyond pre-KPgE levels, a doubling of the diversity of Momipites spp. [fossil juglandaceous (walnut family) pollen (Fig. 3I)], and the second early Paleocene warming interval (Figs. 1 and 4). The diversification of Juglandaceae taxa with small, winged seeds to later taxa with larger, wingless seeds is hypothesized to reflect a transition from wind to animal transport (39). This hypothesis is supported by the close correlation between diversification reflected in fossil juglandaceous pollen and the appearance of several large herbivorous periptychid mammals whose specialized and enlarged premolars are thought to be for hard-object feeding (36, 37). Finally, the appearance of legumes co-occurred with a shift in maximum mammalian body mass and a tentatively recognized short warming pulse. Specifically, two large-bodied mammals appear at ~700 ka post-KPgE (Fig. 4)—the herbivorous multituberculate T. taoensis (~34 kg) and the omnivorous triisodontid archaic ungulate Eoconodon coryphaeus (~47 kg) (Fig. 2, A and B). These data suggest that earliest Paleocene warming pulses may have played an important role in post-KPgE ecosystem recovery, perhaps by facilitating immigration and/or in situ coevolution of flora and fauna.

The transition from an ecosystem characterized by a small-bodied mammalian fauna, “postdisaster” ferns, and low-diversity plant communities to one exhibiting a larger-bodied mammalian fauna and more ecologically and taxonomically complex forests mirrors modern postdisaster ecological successions, but on a much longer time scale (typically 104 to 105 years for recoveries from global mass extinctions versus 101 to 102 years for modern local-regional ecological recoveries) (40). The overall and long-term recovery we observe has recently been described as an aspect of “Earth system succession” (40). This concept proposes that global ecological succession after mass extinctions is intrinsically paced by the interactions of the biosphere and geosphere, both of which may be knocked out of equilibrium (40). The low-diversity, small-bodied mammalian fauna and low-diversity forests dominated by ferns and palms, often indicative of ecological disequilibrium, suggest that a period of ecosystem disequilibrium lasted for up to ~100 ka post-KPgE in our research area. A period of ecosystem recovery followed ~100 to 300 ka post-KPgE when megafloral diversity steadily increased. At ~300 ka post-KPgE, we see several additional signs of ecosystem recovery, including (i) the increase and then plateau of megafloral standing richness; (ii) LMA exceeding pre-KPgE levels; (iii) diversification of Juglandaceae, a potentially energy-rich food source for mammals; and (iv) the first substantial taxonomic diversification, dietary specialization (e.g., increased herbivory), and increase in maximum body mass of mammals (Pu1–Pu2). Finally, spatial niche partitioning, appearance of several additional large (30+ kg) mammals, and expansion of mammalian body mass disparity continue through ~700 ka at the Pu2–Pu3 boundary, all of which are further indications of ecosystem recovery. These changes are correlated with the arrival of plant taxa (e.g., legumes) that would have offered mammals new calorie-dense food sources. Taken together, our record places time estimates on the patterns of biotic recovery in Earth system succession and demonstrates that several aspects of ecosystem recovery occurred within ~300 ka post-KPgE (Fig. 3).

The pattern of warming pulses correlated with biotic change during the earliest Paleocene demonstrates a strong relationship between the biosphere and geosphere. The Deccan Traps of the Indian subcontinent represent repeated and voluminous volcanic eruptions (>106 km3 of magma) during the post-KPgE Earth system succession (6, 7). These eruptions might have induced warming pulses via the release of greenhouse gases (e.g., CO2) (7). Recent work on the timing of these eruptions (6, 7) places ~70% of the total volume within the 300-to-400-ka window roughly coincident with the earliest Paleocene warming pulse(s) observed at Corral Bluffs and the temporally correlated shifts in biotic recovery (Figs. 1 and 4). Although not a feedback of the biosphere-geosphere system, Deccan eruptions likely influenced atmospheric chemistry, in turn shaping Earth system succession and post-KPgE ecosystem recovery (Fig. 4). Detailed records of post–mass extinction biotic recovery, such as the one presented here, will provide a critical framework for predicting ecosystem recovery after mass extinction events, including the one we currently face (41).

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S9

Table S1

References (42123)

Data S1 to S14

References and Notes

Acknowledgments: We thank Norwood Properties, City of Colorado Springs, Waste Management, Aztec Family Raceway, J. Hawkins, J. Hilaire, J. Carner, W. Pendleton, the Bishop family, and H. Kunstle for land access; the State of Colorado, Office of the State Archaeologist, for issuing collection permits; J. Alicia, S. Begin, J. P. Cavigelli, H. Cochard, J. Englehorn, J. Groenke, F. Koether, L. Lacey, A. Lujan, B. Masek, B. Pittman, and N. Toth for preparation of specimens; K. Getty, R. Hess, R. Lavie, S. Milito, Y. Rollot, P. Sullivan, J. Wyman, and L. Taylor for field assistance; F. Cochard, L. Dougan, S. Luallin, R. Wicker, J. Wood, and the USGS National Unmanned Aircraft Systems team for photography; K. MacKenzie and N. Neu-Yagle for collections assistance; and B. Snellgrove for logistics. Funding: Funding was provided by The Lisa Levin Appel Family Foundation, M. Cleworth, Lyda Hill Philanthropies, David B. Jones Foundation, M. L. and S. R. Kneller, T. and K. Ryan, and J. R. Tucker as part of the Denver Museum of Nature & Science (DMNS) No Walls Community Initiative. Author contributions: T.R.L. and I.M.M. led the project. T.R.L. wrote and edited the manuscript in collaboration with I.M.M., A.D.B., K.W., and S.G.B.C. were primary project participants. All authors collected and/or analyzed data and samples, interpreted results, and edited the manuscript. Competing interests: None declared. Data and materials availability: Fossil specimens and pollen slides are all deposited at the DMNS. All data are available in the supplementary materials.

Stay Connected to Science


Navigate This Article