Low dinosaur biodiversity in central China 2 million years prior to the end-Cretaceous mass extinction

Significance

We collected over 1,000 dinosaur eggshell samples from an ∼150-m-thick stratigraphically continuous fossil-rich sequence in the Shanyang Basin of central China, which is one of the most abundant dinosaur records from a Late Cretaceous sequence. We use biostratigraphy of dinosaurs and Bemalambda, magnetostratigraphy, and cyclostratigraphy from orbital cycles to establish a geochronological framework of the dinosaur fossils with a high resolution of 100,000 y. Our results demonstrate low dinosaur biodiversity during the last 2 million y of the Cretaceous, and those data indicate a decline in dinosaur biodiversity millions of years before the Cretaceous/Paleogene boundary. The end-Cretaceous catastrophic events, such as the Deccan Traps and bolide impact, probably acted on an already vulnerable ecosystem and led to nonavian dinosaur extinction.

Abstract

Whether or not nonavian dinosaur biodiversity declined prior to the end-Cretaceous mass extinction remains controversial as the result of sampling biases in the fossil record, differences in the analytical approaches used, and the rarity of high-precision geochronological dating of dinosaur fossils. Using magnetostratigraphy, cyclostratigraphy, and biostratigraphy, we establish a high-resolution geochronological framework for the fossil-rich Late Cretaceous sedimentary sequence in the Shanyang Basin of central China. We have found only three dinosaurian eggshell taxa (Macroolithus yaotunensis, Elongatoolithus elongatus, and Stromatoolithus pinglingensis) representing two clades (Oviraptoridae and Hadrosauridae) in sediments deposited between ∼68.2 and ∼66.4 million y ago, indicating sustained low dinosaur biodiversity, and that assessment is consistent with the known skeletal remains in the Shanyang and surrounding basins of central China. Along with the dinosaur eggshell records from eastern and southern China, we find a decline in dinosaur biodiversity from the Campanian to the Maastrichtian. Our results support a long-term decline in global dinosaur biodiversity prior to 66 million y ago, which likely set the stage for the end-Cretaceous nonavian dinosaur mass extinction.

The demise of nonavian dinosaurs at the end of the Cretaceous is a large component of one of the most severe mass extinctions of the Phanerozoic. The impact of an asteroid roughly 10 km in diameter in the Yucatán Peninsula (Chicxulub, Mexico) and the resulting environmental destruction is a widely accepted causal mechanism for nonavian dinosaur extinction (1). Despite that consensus, there are extensive ongoing disagreements about whether the dinosaur extinction was geologically abrupt (coinciding with the impact) or more gradual, occurring over millions of years (2–10). Two hypotheses have been proposed to reconstruct the tempo and mode of nonavian dinosaur extinction at the end of the Cretaceous. One uses various analytical techniques applied to fossil record data to suggest that there is no evidence for a decline in nonavian dinosaur diversity prior to their extinction 66 million y ago (Ma) (2, 5). The other hypothesis supports a decline in diversity over the much longer timescale of hundreds of thousands to even tens of millions of years (3, 7, 9). That proposed decline in dinosaur diversity has been linked to various abiotic and biotic factors, including ecological competition and interdependence among species (3, 7). The latter hypothesis suggests that the long-term decline in dinosaur diversity reduced their ability as a group to adapt to drastic environmental changes during the latest Cretaceous, and as a result, they were unable to respond sufficiently to recover from the disruptions caused by the bolide impact (3). There is no consensus on whether or not dinosaurs declined prior to the Cretaceous-Paleogene boundary (KPB) because of biases in the dinosaur fossil record and differences in analytical methods used. This conflict limits our broader understanding of the mechanisms of both dinosaur diversification and their extinction.

The known fossils of Late Cretaceous dinosaurs are distributed primarily in North America and East Asia (6, 7, 11). Currently, only the Hell Creek Formation of the North American Western Interior Basin provides a well-sampled and relatively stratigraphically continuous record of dinosaurs during the final million years of the Cretaceous, and it documents the dinosaur diversity before the mass extinction (8, 12). However, the diversity of dinosaurs and their evolutionary patterns on the interior coastal plain in North America may be unique and not representative of the global diversity at the close of the Cretaceous and the interrelated pattern of extinction. To address these issues on a global scale, extensive records from the same time frame are required from outside of North America.

Variation in the diversity of dinosaurs in the Late Cretaceous is thought to be linked to abiotic drivers, such as global climatic fluctuations (7), sea level changes (3, 13–15), plate tectonic movements (16), and Deccan Traps eruptions (17). It has been suggested that the diversity dynamics of Late Cretaceous dinosaurs in North America are intrinsically related to the retreat of the Western Interior Seaway through: 1) increased faunal mixing and a reduction in allopatric speciation; and 2) a reduction in adequate conditions for fossil preservation (4, 6). Despite those proposals for regional changes, the identification of the global drivers impacting dinosaur diversification dynamics is not complete.

In this study, we apply data from magnetostratigraphy, cyclostratigraphy, and biostratigraphy to establish a high-precision geochronological framework for the abundant dinosaur eggshells distributed through the Late Cretaceous sediments of the Shanyang Basin in East Qinling, central China (Fig. 1). By comparing accurately and precisely dated dinosaur fossil records in central China and North America, we can better understand global-scale changes in their biodiversity through the Late Cretaceous.

Fig. 1.

Location and geology of the Shanyang Basin in East Qinling, central China (A). Four studied stratigraphic sections (Sigou, Dongbeigou, Juanlingcao, and Niupanggou) are shown with red lines. Most dinosaur eggshells in this study were found in the layers from site a to site b within the Niupanggou section (B). The lithology refers to Fig. 2. The topographic maps are based on the digital elevation database from https://srtm.csi.cgiar.org and https://www.google.com/maps.

Geological Setting and Sampling

The Shanyang Basin is located in the East Qinling area of central China (Fig. 1A), underlain by Paleozoic rocks, including the Upper Devonian Liuling Formation, cropping out in the northern part of the basin and the Upper Carboniferous Tiechangpu Formation exposed along the southern margin (Fig. 1A). The Cretaceous–Cenozoic Shanyang Basin sedimentary sequence unconformably overlies these Paleozoic rocks and can be divided into three formations, including the Upper Cretaceous Shanyang Formation, the Paleocene Juanling Formation, and the Oligocene Guanyinsi Formation (18) (Fig. 1A). The Shanyang Formation, consisting of brownish-red thick-bedded conglomerates and sandstones, is the primary sedimentary sequence distributed across most of this basin. The Juanling Formation consists of intercalated yellowish-red argillaceous siltstones and gravelly sandy mudstones, distributed mainly in the eastern part of the basin. The Guanyinsi Formation, comprised of reddish conglomerates, is confined to a limited area on the southern margin of the basin. It is believed that the onset of basin deposition was controlled by normal faulting along the Shanyang fault, which bounds the southern margin of the Shanyang Basin. The stress field of the Shanyang Basin changed from extension to compression in the early Paleogene, leading to the cessation of lacustrine sedimentation in the basin (18).

A total of 3,538 oriented paleomagnetic samples from 1,180 stratigraphic levels, 5,466 unoriented samples with a stratigraphic resolution of ∼5 cm, and more than 1,000 well-preserved dinosaur eggshells (several complete dinosaur eggs) from 44 stratigraphic levels were collected across four sections of the Shanyang Basin (Materials and Methods and Fig. 1).

Magnetostratigraphy and Cyclostratigraphy

A total of 558 of the 919 measured paleomagnetic specimens yielded reliable characteristic remanent magnetization (ChRM) directions (SI Appendix, Tables S5–S8). The ChRMs proved to be primary by the conglomerate test and reversals test (SI Appendix), and therefore, they can be utilized to establish a magnetostratigraphic framework for the fossiliferous stratigraphic sequence in the Shanyang Basin (Fig. 2). The stratigraphic sections are further constrained through biochronology, using the first and last occurrences of the Danian pantodont mammal Bemalambda and Cretaceous dinosaur fossils (Fig. 2A). Magnetostratigraphic analysis shows that the Shanyang Basin sedimentary sequence spans part of the Late Cretaceous and Paleogene from Chron C32n.1n to the lower Chron C28n in the geomagnetic polarity timescale (GPTS 2020) (19), roughly equivalent to a ∼7.1-million-y duration spanning from ∼71.7 to ∼64.6 Ma (Materials and Methods and Fig. 2).

Fig. 2.

Lithostratigraphy, biostratigraphy, and magnetostratigraphy of the Shanyang Basin sedimentary sequence. (A−C) Juanlingcao section; (D−F) Sigou section; (G−I) Dongbeigou section. GPTS (J) is the geomagnetic polarity timescale (GPTS 2020) (19). VGP latitudes are for sample ChRM directions converted to VGPs. Positive (northerly) and negative (southerly) VGP latitudes correspond to normal and reverse polarities, delineated in the polarity column by filled and open bars, respectively. Green open circles are accepted sample data with maximum angular deviation (MAD) values of 15° or less (SI Appendix). The proposed correlation constrains the Shanyang Basin sedimentary sequence to an interval from ∼71.7 Ma to ∼64.6 Ma, based on the occurrences of the Danian fossil Bemalambda and the Cretaceous dinosaur fossils (labeled open triangles). The thick conglomerate layer (labeled solid triangles), together with the overlying pelitic siltstone embedded a mass of reduction spots, are performed as a marker layer to correlate different sections.

Based on mean sediment accumulation rates constrained by the magnetostratigraphic framework (SI Appendix, Fig. S8) and comparisons between the frequency ratios of proxies and ratios of Milankovitch frequencies, 38 individual 100,000-y eccentricity cycles were counted (SI Appendix, Fig. S13). Our tuning indicates a ∼3.9-million-y duration for the interval of the ∼300-m-thick (from stratigraphic level 16.0 m to level 316.5 m) Sigou section, providing a high-resolution astronomical timescale for the dinosaur fossils found in the Shanyang Basin (Fig. 3 and SI Appendix, Fig. S13).

Fig. 3.

Ages of the dinosaur eggshells and bones in the Shanyang Basin. The integrated lithology (A and B) refers mainly to the Juanlingcao section for its integrity and continuity. The high-resolution magnetic susceptibility data (D) were acquired from the Sigou section. The labels e7–e33 in D represent the 100,000-y short eccentricity cycles extracted with Gaussian band-pass filters using the software of Analyseries 2.0.8 (52). The ages in red (black) are the astronomical timescale (GPTS) (C). Dinosaur bone fossils in E refer to Xue et al. (18). Layer I: dinosaur species undetermined. Layer II: Shanyangosaurus niupanggouensis. Layer III: Shantungosaurs cf. giganteus, Tyrannosauridae, Sauropoda; Layer IV: dinosaur species undetermined. (SI Appendix). The relative thicknesses of the Sigou section and Niupanggou section are converted to the thickness of the integrated section based on their polarities and lithology (e.g., conglomerate layers). The δ¹⁸O values (F and G) are from the tropical Pacific (ODP 1209, ODP 1210B) (42, 55), Walvis Ridge in the South Atlantic (ODP 1262) (43), and Boreal Chalk Sea in Denmark (Stevns-1 core) (56). The δ¹⁸O values of the ODP cores were measured from benthic foraminifera. The δ¹⁸O values of the Stevns-1 cores were measured from the bulk carbonate of Stevns-1. Abbreviations: KPB, Cretaceous-Paleogene boundary; MMWE, Middle Maastrichtian warming event; LMWE, Late Maastrichtian warming event; and DT, Deccan Traps.

Dinosaur Fossils and Taphonomic Environments in the Shanyang Basin

Dinosaur eggshells, eggs, and bones have been recovered chiefly in the eastern part of the Shanyang Basin (Fig. 1). Since their first discovery in the 1980s, Xue et al. (18) identified the Shanyang Formation dinosaur eggs as belonging to the eggshell taxa (ootaxa) Macroolithus yaotunensis, Elongatoolithus elongatus, Lepidetoolithus guofenlouensis, Ovaloolithus cf. chinkangkouensis, Shanyangoolithus shanyangensis, and Paraspheroolithus lamelliformae. Later, Zhao et al. (20) revised L. guofenlouensis and P. lamelliformae by synonymizing them with M. yaotunensis and Stromatoolithus pinglingensis, respectively. They also identified S. shanyangensis as a turtle egg and questioned the validity of the oospecies O. cf. chinkangkouensis (20). Our study suggests that the main ootaxa in the Shanyang Basin are M. yaotunensis, E. elongatus, and S. pinglingensis (Figs. 1B and 3E and SI Appendix). The lowest layer that contains dinosaur eggshells (M. yaotunensis) and bones is at the stratigraphic level of ∼360 m (Fig. 3), ∼4.5 million y prior to the KPB. The first appearance of the dinosaur eggshells of E. elongatus in the Shanyang Basin is ∼68.2 Ma (Fig. 3). All types of dinosaur eggshells occurring in the Shanyang Basin are present from the stratigraphic level of ∼500 m (∼67.8 Ma) (Fig. 3). Subsequently, the dinosaur fossil occurrences decrease sharply from the stratigraphic sequence of ∼569 m (∼66.8 Ma) (Fig. 3), with the stratigraphically highest occurrences deposited ∼0.4 million y before the KPB.

The lithology of the stratigraphic interval 0 to 320 m is matrix-supported polymictic conglomerates with cross-bedding coarse- and medium-grained sandstone interlayers, which are interpreted as alluvial fan deposits. Red fine-grained argillaceous siltstones and interbedded green medium-grained sandstones with horizontal bedding structures dominate the stratigraphic interval between 320 and 460 m, which are considered to have been deposited in a shallow lake system. A few eggshells of M. yaotunensis and plentiful calcareous nodules were discovered within these sediments. The lithology of the interval of 460 to 570 m is fine-grained red argillaceous sandstones with green medium-grained pebbly sandstone interlayers (Figs. 3 and 4). These sedimentary sequences are considered to have formed in meandering river systems as indicated by the cross- and parallel-stratified sandstones representing channel fills and the fine sediments of the floodplain sediments. Abundant stratigraphically continuous dinosaur eggshells and several dinosaur bone fossils were collected in these meandering river deposits. Dominant clast-supported polymictic conglomerates and medium-grained sandstone interlayers with scour structures and cross-bedding structures in the stratigraphic interval of 570 to 632 m indicate the development of a braided river system. The lithology, sedimentary structures, and Bemalambda fossils present in the stratigraphic range of 632 to 970 m point to the occurrence of meandering river facies in the early Paleogene (Fig. 4).

Fig. 4.

Sedimentary facies in the Juanlingcao–Niupanggou area. Similar lithology and sedimentary structures were observed in the Juanlingcao section and Niupanggou section. The tectonic evolution of the Shanyang Basin refers to Yu and Li (57). The SAR (sediment accumulation rate) in the different stages was calculated by the age model and the thickness of the sediments (SI Appendix, Fig. S8). The warm and cool climate intervals refer to Fig. 3.

The sharp corners, well-preserved surface ornamentation, and the nonuniform size of most eggshells (SI Appendix, Figs. S16 and S17) indicate that the eggshells were buried in situ or transported only a short distance. The stratigraphically continuous distribution of well-preserved eggshells between the stratigraphic levels of 460 and 570 m shows that the meandering river system and the climate during this period were suitable for the preservation of dinosaur fossils (Fig. 5). Though the number of stratigraphic levels containing dinosaur fossils in the braided river system is not as great as that in this meandering river system, there are also three dinosaurian eggshell taxa in the braided river deposits. Hence, we conclude that the pattern of eggshell taxa distribution observed in the Shanyang Basin has biological significance rather than representing a taphonomic bias.

Fig. 5.

Paleogeography of the Shanyang Basin in the stage of the Maastrichtian meandering river system. The depositional center is in the eastern part of the basin (18), and its sediment provenances are mainly from the southern and northwestern denudation areas (57). Most dinosaur eggshells buried in situ were discovered in the eastern part of the basin (e.g., the Niupanggou area), indicating that the dinosaurs primarily laid their eggs in this area. The slow subsidence of the hanging wall block and the developed meandering river system provide an excellent taphonomic environment for dinosaur egg preservation.

Discussion

Dinosaurs in Decline Before the KPB in Central China.

The eggshell taxon M. yaotunensis was laid by oviraptors based on adult–egg–embryo associations (21). Shared macro- and microstructural aspects of M. yaotunensis indicate that E. elongatus likely is related to theropod dinosaurs as well, and provides further support for the relationship between Elongatoolithus and oviraptors (22, 23). Stromatoolithus pinglingensis has been considered to be derived from hadrosaurs (24, 25). These hypothesized eggshell producers are consistent with known dinosaur bones from the basin, and that diversity includes Shanyangosaurus niupanggouensis (a small coelurosaurian theropod), an unidentified tyrannosaurid, Shantungosaurus cf. giganteus (a hadrosaur), and a few specimens of sauropods (18) (Fig. 3). Moreover, a series of comparable fossil assemblages comprising dinosaur eggshells and bones have been found in several basins adjacent to the Shanyang Basin (18, 26–29) (Fig. 6B), and they demonstrate extensive movement and communication of dinosaur populations among these basins in East Qinling, and even between northern and southern China during the Late Cretaceous (18). The basins in East Qinling may have functioned as a single wider basin during the Late Cretaceous resulting from the planation of the Qinling orogenic belt during the Cretaceous (18, 30), which led the Shanyang Basin dinosaurian fauna not being isolated from other parts of China.

Fig. 6.

The Late Cretaceous dinosaur bones and eggshells through time (A) in the East Qinling basins (B) and the global distribution of the Late Cretaceous dinosaur occurrences (C). The materials and data in A are from the basins in the East Qinling region (B) (18, 26–29). The ranges of the basins in East Qinling (B) are the coverage area of the sediments currently seen. The fossil ages and the ranges of the same basin are presented in the same color (e.g., the Shanyang Basin is presented in black). The global dataset of the Late Cretaceous dinosaur refers to Condamine et al. (7). The topographic maps (B) are based on the digital elevation database from https://www.google.com/maps. The plate tectonic map 66.0 Ma is produced by the ODSN Plate Tectonic Reconstruction Service (https://www.odsn.de/odsn/services/paleomap/paleomap.html). See SI Appendix, Table S12 for the source of dinosaur images. Abbreviations: Cen., Cenomanian; Tur., Turonian; Con., Coniacian; San., Santonian; Cam., Campanian; and Maa., Maastrichtian.

Analysis of all of the dinosaur records in East Qinling (Fig. 6 A and B) shows that eight clades of dinosaurs are present in East Qinling during the Campanian (26–29), but only three families (Oviraptoridae, Hadrosauridae, and Tyrannosauridae), and a few specimens of sauropods (18) are known from the late Maastrichtian of the region (Fig. 6A). Despite lacking precise chronological constraints, abundant dinosaur egg deposits found in southern and eastern China represent a comparable variation in (low) diversity from the Campanian to the Maastrichtian (31), and only elongatoolithids and S. pinglingensis were found in the late Maastrichtian sediments of the Nanxiong Basin in southern China (32). These discoveries suggest a long-term decline in diversity among dinosaurian lineages from the Campanian to the late Maastrichtian (Fig. 6A). In particular, our data point to a low level of dinosaur diversity in the Shanyang Basin sustained for at least 2 million y prior to the end-Cretaceous mass extinction.

Our dinosaur data from the Shanyang Basin are largely congruent with recently published global diversity dynamics models of Late Cretaceous dinosaurs (7). Those published data based precisely on the North American record (Fig. 6C) demonstrate that carnivorous theropod groups (e.g., Dromaeosauridae, Troodontidae, and Tyrannosauridae) were in decline starting near the Campanian–Maastrichtian transition (∼72 Ma), and that hadrosaurids as a formerly dominant ornithischian clade declined at a slower rate than that of the theropod groups (7). Additionally, the Mesozoic evolutionary dynamics of speciation and those of the extinction of dinosaurs reconstructed with Bayesian phylogenetic methods support a long-term decline across all dinosaur lineages, except for the Hadrosauriformes and Ceratopsidae starting about 40 million y prior to the KPB (3). Oviraptorosaurs are an herbivorous group of feathered bird-like theropod dinosaurs mainly found in East Asia and North America (33, 34), with more than 35 known genera. While that diversity thrived in the Campanian, their diversity decreased starting around 72 Ma (33). When examined in a larger context (Fig. 6C), it seems that the steady low dinosaur diversity during the late Maastrichtian of the Shanyang Basin can be interpreted as the result of a long-term decline among dinosaurs over many millions of years.

In contrast, several studies do not support a long-term global decline of dinosaur diversity prior to the KPB (2, 5, 35). This apparent conflict can be ascribed to sampling biases in the recovery of dinosaur fossils and differences in the application of analytical methodologies (5, 10). Moreover, the rarity of high-precision dating for dinosaur fossils is another barrier to fully resolving issues related to determining the tempo and causes of the end-Cretaceous dinosaur extinction. To resolve these issues, it is essential to increase the density and spatiotemporal scope of taxon sampling in combination with well-dated, regional-level analyses. Our study in Asia of abundant and geochronologically dated fossils is a major step in that direction.

High-resolution biostratigraphic analysis of 287 dinosaurian macrofossils and 138 bonebeds in the Edmonton Group (Upper Cretaceous) demonstrated the diversity of ornithischians and theropods dropped sharply by ∼68 Ma, long before the end of the Cretaceous (36). Dean et al. (6) used an increased temporal resolution method to investigate the late Cretaceous diversity dynamics of dinosaurs in the Western Interior of North America, and they pinpointed the timing of the Maastrichtian diversity decline to between 68 Ma and 66 Ma. That temporal interval largely overlaps with our high-resolution dated dinosaur eggshell record from the Shanyang Basin, and points to a lowered diversity and overall decline among dinosaurs on a global scale prior to the KPB. In addition, the consistent diversity pattern derived from dinosaur eggshells and bones suggests that eggshells are a good proxy for dinosaur diversity at family/clade level in a fauna.

Factors in Dinosaur Diversification.

Our dinosaur data and previously published dinosaur diversity dynamics (7) indicate that hadrosaurs and oviraptorosaurs were probably the dominant dinosaurs among populations in the Northern Hemisphere during the Maastrichtian. The complex tooth batteries capable of crushing, grinding, and shearing helped hadrosaurs to consume a broad herbivorous diet (37, 38), and those generalist herbivorous diets may have been crucial for hadrosaurs to outcompete other herbivores (e.g., ankylosaurs, ceratopsians) occupying similar ecological niches (7). Based on the presence of shearing adaptations in the mandible of Chirostenotes pergracilis (Oviraptorosauria) and the occurrence of gastrolith masses in other related species, the clade most likely had a largely herbivorous diet as well (39). Along with their nesting and brooding behavior (21), oviraptorosaurs may have been more resistant to climate and habitat changes (4) during the Late Cretaceous. Analyses of Late Cretaceous ecosystem networks demonstrate that declines in a highly connected group (e.g., ankylosaurs) (7, 40) could have had consequences throughout the entire food web, and that may have made dinosaurian faunas more susceptible to environmental changes.

Although the dinosaur fauna in the Shanyang Basin maintained a low diversity during the last 2 million y of the Cretaceous, we can still analyze the abiotic factors relevant to dinosaur diversification dynamics from the relationship between the distribution of fossiliferous layers and climate changes during that period. The global mean temperature experienced a long-term decline from the Cenomanian to the Maastrichtian (41), but marine and terrestrial sediments record multiple short-term warming and cooling climatic events from the latest Cretaceous to the earliest Paleogene (42–45) (Fig. 3). The oldest record of dinosaur fossils in the Shanyang Basin is constrained to 70.5 Ma (Fig. 3), coincident with the transition from an alluvial fan system to shallow lake depositional environments. However, the suitable habitat for dinosaurs during this period was limited because of the constrained range of the basin in its initial development. The middle Maastrichtian warming event (MMWE) starting at 69.5 Ma is characterized by an ∼2 to 4 °C warming of marine bottom and surface waters, and an ∼6 °C warming in terrestrial air temperatures (46) (Fig. 3 F and G). The MMWE was thought to be caused by the intense volcanic activity on the Ninety East Ridge (47), which induced poleward-shifted westerlies over the Asian continent and enhanced the intensity of the hydrological cycle (45). During the MMWE, the strengthening of water circulation may have led to increased rainfall, which is beneficial for the development of the lake system and the basin.

The MMWE was followed by climatic cooling with a decrease in marine temperatures by ∼1 to 3 °C and terrestrial temperatures by ∼3 to 5 °C ending ∼66.5 Ma (42, 46). Most dinosaur fossils found in the Shanyang Basin were deposited during this cooling period (Fig. 3 E–G). Notably, the number of dinosaur fossils decreases and the gravel content of the sediments increases starting at ∼66.8 Ma in the Shanyang Basin (Fig. 3), which is consistent with the transition from meandering river to braided river at ∼66.9 Ma (Figs. 3 and 4). This change may have resulted from the enhanced intensity of the hydrological cycle and chemical weathering driven by a long-term gradual warming starting after 66.8 Ma (43). The dinosaurs of the Shanyang Basin either immigrated away or died out at the onset of the late Maastrichtian warming event (LMWE) (2 to 4 °C warming episode), which occurred about 300,000 y before the KPB (43). The Deccan Traps volcanism erupted quasi-continuously from 66.413 Ma to 65.422 Ma (17), and that volcanism with its release of large amounts of climate-modifying gases (CO₂, CH₄, and SO₂) was thought to be the most likely driver of the LMWE (43, 48). Additionally, we found more complete dinosaur eggs in the braided river deposits of the upper part of member 2 of the Shanyang Formation. This occurrence may imply that the climate or sedimentary environment during this period was less suitable for dinosaurs to hatch their eggs, a problem which could have impacted the size and even the diversity of dinosaur populations. Nonetheless, this hypothesis and preservation pattern requires further investigation.

In conclusion, our analyses of the sedimentary facies and taphonomic environments reveal that the stratigraphically continuous eggshells of the Shanyang Basin are largely well-preserved and were buried in situ. We propose that the demonstrated low dinosaur biodiversity in central China during the late Maastrichtian is representative of a long-term global decline in dinosaur diversity that set the stage for the eventual extinction of nonavian dinosaurs. Our results raise the prospect of a greater integration of the understanding of global dinosaurian extinction patterns. Additionally, this study highlights the importance of obtaining accurate and precise geochronological age constraints in paleobiodiversity studies, and we advocate for additional work to build a more detailed geochronological framework for examining variation in global dinosaur biodiversity variation through deep time.

Materials and Methods

Samples and Fossils.

We performed magnetostratigraphic studies of four representative stratigraphic sections, namely the Juanlingcao, Dongbeigou, Sigou, and Niupanggou sections in the Shanyang Basin, central China (Fig. 1A). The Juanlingcao section contains a complete sedimentary sequence of the Shanyang Basin. The lower and upper parts of the Shanyang Basin sedimentary sequence are exposed in the Dongbeigou and Sigou sections, respectively (Fig. 1A). Three parallel samples with independent orientations were collected from each stratigraphic level. A total of 3,538 paleomagnetic samples from 1,180 levels were collected across these four sections (Fig. 1A and SI Appendix). The oriented paleomagnetic samples were drilled using a portable gasoline-drilling machine. Cores were oriented in the field with magnetic and solar compasses (SI Appendix). Dinosaurian and mammalian fossils provided biostratigraphic tie points for the magnetostratigraphic sequence. A total of 5,466 unoriented samples were collected with a resolution of ∼5 cm from the stratigraphic levels between 16 and 316.5 m in the Sigou section for cyclostratigraphic study.

We systematically collected more than 1,000 well-preserved dinosaur eggshells and several complete and incomplete dinosaur eggs from 44 stratigraphic levels of sediments in the Niupanggou section (Fig. 1B) and Juanlingcao sections (SI Appendix, Fig. S14). The dinosaur eggshells and eggs were buried in situ or transported only a short distance (SI Appendix, Figs. S15–S17), and therefore, they are indicative of the original stratigraphic context of the dinosaurs and their habitats in the Shanyang Basin.

Paleomagnetism.

Measurements of the natural remanent magnetization and the remanences after thermal demagnetization were performed by using a three-axis cryogenic magnetometer (2G 760) in a magnetically shielded room (residual field < 300 nT). A total of 919 specimens from 1,180 stratigraphic levels were subjected to progressive thermal demagnetization (22 steps) up to a maximum temperature of 690 °C, with intervals of 25 to 50 °C below 610 °C and 10 °C above 610 °C, using thermal demagnetizers (MMTD80A and ASC TD−48) with residual magnetic fields of less than 20 nT. The thermal demagnetization results show that the secondary magnetic component, probably of viscous remanent magnetization, was removed at lower than 300 °C. The high-stability ChRMs were isolated from the demagnetization steps higher than 300 °C (SI Appendix). The vector endpoint demagnetization diagrams (49) and the least-squares fitting technique (50) were used to compute the demagnetization results and the principal component directions. A total of 558 specimens from the four sections (Fig. 1 and SI Appendix, Tables S5–S8) of the Shanyang Basin yielded reliable ChRM directions. A plot of the virtual geomagnetic pole (VGP) latitudes with thickness in the Juanlingcao section reveals nine magnetic polarity zones designated from the top down as N1 to N5 and R1 to R4 (Fig. 2C).

The specimens near the sedimentary marker layer at the stratigraphic level of ∼640 m, characterized by an ∼20-m-thick conglomerate layer (with the overlying pelitic siltstone embedded lots of green reduction spots), record a reversed polarity (R2) (Fig. 2B). Abundant dinosaur eggshells and bones were found beneath this marker layer, and only early Paleogene (Danian) mammalian fossils [namely Bemalamda (51)] have been found above it (18). Hence, the reversed polarity R2 (Fig. 2C) should be correlated to Chron C29r as it is the reversed polarity interval encompassing the KPB. With that result, polarities N3 (485.4 to 614.8 m), R3 (460.9 to 485.4 m), and N4 (431.5 to 460.9 m), respectively, correspond to Chrons C30n, C30r, and C31n of the GPTS (19) (Fig. 2 C and J). The specimens taken from the sandstone interlayers of the lower conglomerate strata record normal polarity N5 (Fig. 2B), and no paleomagnetic sample was taken from the upper layer (∼70 to 320 m) of this conglomerate unit. However, a sandstone specimen taken just above the underlying conglomerate unit documents a normal polarity (Fig. 2B), and that probably belongs to the normal polarity N5 of the Juanlingcao section (in light of the rapid accumulation of the underlying conglomerate strata during the early depositional stage of the basin formation). Hence, polarity N5 (Fig. 2C) is most likely correlated with Chron C32n.1n (71.689 to 71.449 Ma). It is reasonable to correlate polarities N2, R1, and N1 with Chrons C29n, C28r, and C28n (Figs. 2 C and J), respectively, because there is no or only a brief hiatus in the Juanlingcao section. Consequently, the age of cessation of deposition in the Shanyang Basin is close to the C28r−C28n boundary at ∼64.6 Ma. The polarities of the Dongbeigou, Sigou, and Niupanggou sections can be correlated to the GPTS (19) as well, showing a consistent pattern with the Juanlingcao section in both their lithologies and magnetic polarities (Fig. 2 and SI Appendix, Fig. S7).

Cyclostratigraphy.

Magnetic susceptibilities were measured using a Bartington MS2 Magnetic Susceptibility System. The magnetic susceptibility series were prewhitened in Kaleidagraph software by subtracting 25% weighted averages to remove long-term trends. Constrained by the magnetobiochronology and the calculated sediment accumulation rates, the dominant spectral components (short and long eccentricity cycles, obliquity, and precession cycles) were extracted using Gauss band-pass filtering in the software AnalySeries 2.0.8 (52). The power spectra of the untuned and tuned data were analyzed using the 2π-MultiTaper Method (MTM) in the SSA-MTM Toolkit with robust red noise models set at mean, 90%, 95%, and 99% confidence levels (53). An astronomical timescale was established by tuning to 100-kyr short eccentricity cycles interpreted in the sedimentary records to the La2010 astronomical model (54).

Dinosaur Eggshells.

All eggshells were ultrasonically washed, and those with both the internal and external surfaces intact were selected. Histological thin sections were made from the preserved hard tissues section to examine microstructural features. The eggshell sections were examined under normal and orthogonal polarized light using a polarizing microscope for parataxonomic identification. See the detailed information of dinosaur eggshells in SI Appendix, Figs. S14–S21 and Table S10.

Data, Materials, and Software Availability

The measured paleomagnetic data in this study can be downloaded at https://doi.org/10.5281/zenodo.7025078 (58), and the paleomagnetic results are provided in SI Appendix, Tables S5–S8. The magnetic susceptibility data are included in SI Appendix, Dataset S1. The dinosaur eggshells and eggs are deposited in the Department of Specimen Collection of the Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, and are available upon request.