Giant gar from directly above the Cretaceous–Palaeogene boundary suggests healthy freshwater ecosystems existed within thousands of years of the asteroid impact

Abstract

The Cretaceous–Palaeogene (K–Pg) mass extinction was responsible for the destruction of global ecosystems and loss of approximately three-quarters of species diversity 66 million years ago. Large-bodied land vertebrates suffered high extinction rates, whereas small-bodied vertebrates living in freshwater ecosystems were buffered from the worst effects. Here, we report a new species of large-bodied (1.4–1.5 m) gar based on a complete skeleton from the Williston Basin of North America. The new species was recovered 18 cm above the K–Pg boundary, making it one of the oldest articulated vertebrate fossils from the Cenozoic. The presence of this freshwater macropredator approximately 1.5–2.5 thousand years after the asteroid impact suggests the rapid recovery and reassembly of North American freshwater food webs and ecosystems after the mass extinction.

1. Introduction

The Cretaceous–Palaeogene (K–Pg) extinction is the most recent mass extinction in Earth's history and instigated a complete restructuring of terrestrial ecosystems to mammal-dominated communities [1,2]. This event was responsible for the loss of 70–80% of biodiversity [3–5], including the infamous demise of the non-avian dinosaurs [6,7]. At the same time, this extinction opened numerous ecological opportunities and is associated with adaptive radiations and prolonged diversifications in major extant lineages like neoavian birds [8–10], placental mammals [1,11], snakes [12,13], spiny-rayed fishes [14,15] and flowering plants [16]. Many lineages also experienced a pronounced decrease in average body size across the K–Pg boundary, a phenomenon known as the Lilliput effect [17–19].

The Lilliput effect is difficult to assess for any extinction because it requires an excellent fossil record that brackets the extinction event [19]. For this reason, the Lilliput effect has been best documented in marine groups like planktic foraminifera [20], mollusks [21] and lamniform sharks [22], but has also been reported in terrestrial environments [23]. While the incomplete terrestrial fossil record for many clades currently precludes an analysis of the Lilliput effect across end-Cretaceous ecosystems, it is well documented that large-bodied species go extinct at the K–Pg (e.g. [24–27]), leaving the earliest Palaeocene largely devoid of large-bodied animals. Recent discoveries document the tempo of increasing body size in the aftermath of the end-Cretaceous extinction [28]. Because many large-bodied animals require established food webs, body size is an important variable in assessing overall ecosystem health in the wake of mass disasters like the K–Pg event [29].

Here, we describe a nearly complete skeleton (figures 1 and 2) that represents a new, large-bodied species of fish in the clade Lepisosteidae. This lineage includes the seven extant species of gars and is characterized by low species diversity and morphological conservatism [30–33]. The new species, †Atractosteus grandei sp. nov., is among the largest known gars and holosteans, reaching a length of approximately 1.5 m. Yet, †A. grandei was recovered 18 cm above the K–Pg boundary and lived an estimated 1500–2500 years after the asteroid impact (see electronic supplementary material, table S1 for list of ages using different age models). The exceptional size of †A. grandei contrasts with the small size of most terrestrial vertebrates in the immediate aftermath of the K–Pg mass extinction and thus provides insights into the tempo of recovery and reassembly of Cenozoic North American freshwater ecosystems.

Figure 1.

Anatomy of †Atractosteus grandei sp. nov. Skull in (a) dorsal, (b) ventral and (c) lateral views. Inset shows size compared to a 1.85 m tall man.

Figure 2.

Size and phylogenetic relationships of †Atractosteus grandei sp. nov. (a) Linear regression of HL against SL in a sample of gars, with red star indicating the position on the regression length corresponding to the length of the skull of †A. grandei sp. nov. (b) Time-calibrated maximum clade credibility tree from analysis of the morphological matrix in BEAST 2.6.6. Red line denotes the position of the K–Pg boundary/recovery period. Ma = millions of years.

2. Methods

(a) . Computed tomography

The skull of the holotype of †Atractosteus grandei, Denver Museum of Nature & Science (DMNH) EPV.128289, was scanned at the High-Resolution X-ray CT Facility of the University of Texas by Matthew Colbert. Scan parameters were: North Star Imaging scanner; Fein Focus High Power source, 200 kV, 0.17 mA, brass filter, source to object 378.661 mm, source to detector 733.437 mm, isometric voxel size = 103.5 µm, total slices = 3510. Resolution of the data was to 0.1035 mm. Sixteen-bit TIFF stacks of raw scan data were processed in VGStudio MAX v. 3.5.

(b) . Length estimation

In order to compare the dimensions of the new Atractosteus to other lepisosteids, we collected measurements from the literature for the skull lengths of all seven extant and five crown-group fossil gar species (N = 79 individuals; [34]). Qualitative comparisons of meristic proportions have been used to estimate the sizes of gars (see [34]). A linear regression of head length (HL) versus standard length (SL) for gars produced the following equation relating HL to SL in crown Lepisosteidae: SL = 12.70393 + 3.19394*HL (adjusted R² = 0.9775, F = 3431 on 1, 78 degrees of freedom, p < 2.2 × 10⁻¹⁶). A table with these data is included in the electronic supplementary material.

(c) . Phylogenetic analyses

We assessed the phylogenetic position of †Atractosteus grandei using the matrix of Grande [34] with the updates of [35] and [36]. This matrix consists of 26 taxa coded for 103 morphological characters and is included in the electronic supplementary material, data. Characters were treated as unordered and no implied weighting was used. Parsimony analysis was conducted in TNT v. 1.5 [37] with an initial Wagner search over 10 replicates with default parameters for ratchet, tree fuse, tree drift and sectorial search, followed by a round of traditional bisection–reconnection branch swapping over 100 000 trees. We also conducted a Bayesian analysis using the Fossilized Birth–Death model [38] and a relaxed lognormal clock in BEAST 2.6.6 [39], with dates based on the literature [34,36]. The analysis was run for 10 million generations with a 1.0 × 10⁷ pre-burnin, and convergence was checked using v. Tracer 1.7.2 [40]. Trees were pooled and combined into maximum clade credibility topology in TreeAnnotator 2.6.4 with a 10% burnin [39].

3. Systematic palaeontology

Holostei [41]

Ginglymodi [42]

Lepisosteidae [43]

Atractosteus Rafinesque 1820

†Atractosteus grandei sp. nov.

• (a) Type specimen—DMNH EPV.128289, articulated skull and partial associated skeleton (figure 1).
• (b) Type locality and age—DMNH loc. 7251, Cedar Creek Anticline in the Williston Basin, Mud Buttes study area, Bowman County, North Dakota, USA, 29 km southeast of Marmarth North Dakota (qualified researchers can obtain more detailed locality information from the DMNH); Fort Union Formation, Danian, early Palaeocene; 18 cm above the K–Pg boundary clay and an estimated approximately 1500–2500 years after the K–Pg using different age models (see electronic supplementary material, table S1 for age derived from different age models). The skeleton was found at the base of a 30 cm thick grey, blocky siltstone interpreted to be a ponded water deposit (See electronic supplementary material for more geological and palaeoenvironmental details).
• (c) Etymology—The specific epithet, grandei, in honour of Lance Grande for his contributions to the study of holostean fishes.
• (d) Diagnosis—Distinguished from other Atractosteus species by the following: poorly ornamented skull roof bones; anterior triangular vomerine tooth plate with reduced fangs; vomerine teeth rows mediolaterally restricted to single line; low number of ectopterygoid teeth (nine instead of 12–13 in other species).
• (e) Remarks—†A. grandei is assignable to Atractosteus among lepisosteids based on the medially curved symphyses of the dentary and the shape of the vomerine heads approximately forming a triangle (figure 1; [34]). The skull is robust, resembling the largest two Atractosteus species A. spatula and †A. atrox and an unnamed form from the Cretaceous of Morocco [34,44] rather than A. tristoechus, A. tropicus, †A. simplex, †A. messelensis and †A. falipoui. The skull of †A. grandei also matches the crania of individuals of the largest extant species (A. spatula) in size and was likely approximately 400 mm when complete. This makes †A. grandei one of the largest extinct lepisosteids and holosteans currently recognized [34,45].

The poorly ornamented skull roof is composed of the paired nasals, frontals, parietals, dermopterotics and six extrascapulars. These bones are closely comparable to the same elements in other species of Atractosteus, and the frontals, premaxillae and parietals lack the elongation seen in longirostrine forms like Lepisosteus osseus, †'Lepisosteus’ indicus and †Herreraichthys coahuilensis [34,46]. The premaxillae and frontals are also not heavily abbreviated as in †Masillosteus spp. and †Cuneatus spp. [34,36]. The frontals compose the largest portion of this region and are slightly ornamented with ridges radiating outward from their centres onto the dermopterotics and parietals (figure 1a), as in Atractosteus spp. and all species of Lepisosteus except L. osseus [34]. Ornamentation on the premaxillae, parietals, dermopterotics and extrascapulars also primarily consists of ridges radiating from the ossification centres of each of these bones.

Anteriorly, small, subrectangular elements identifiable as the nasals are present. Laterally, there are seven lacrimomaxillary bones, which become progressively more elongated towards the posterior end of the rostrum. This is the condition in Atractosteus and Lepisosteus [34]. Due to the crushed nature of the posterior skull, the morphology of the lacrimals, suborbitals and other orbital rim bones are largely indiscernible.

Ventrally, the palate is preserved in exquisite detail (figure 1b). The widened vomers are lined with single plates of teeth with parallel margins. Anteriorly, these bear a triangular platform with fangs as in other Atractosteus [34]. However, the fangs are reduced in †A. grandei relative to most other Atractosteus, particularly A. spatula [34]. The ectopterygoids are widened and each bear nine large teeth, compared to 12–13 in other species of Atractosteus [34]. The mandibles show the plesiomorphic condition among lepisosteids retained in Atractosteus but absent in Lepisosteus wherein the dentary symphyses curve medially to contact the opposing element [34]. The basihyal tooth plate is positioned between the mandibles and appears as a rugose, elongated, rectangular element as in other lepisosteids [34]. All teeth exhibit the plicidentine condition.

In addition to the nearly complete articulated skull and lower jaws, the holotype of †A. grandei also includes a partial skeleton consisting of mostly articulated, but incomplete precaudal vertebrae (N = 21), partially dissociated fins, and abundant rhomboid-shaped ganoid scales. The anatomy of the body is difficult to examine, but the vertebrae show the opisthocoelus condition characteristic of the Lepisosteidae [34]. Additional description of †A. grandei is included in the electronic supplementary material.

(a) . Size estimation and phylogenetic analysis

The equation of the regression line relating HL to SL gave an estimated SL of 1.12 m for †A. grandei (HL = 346.5 mm). This SL, which implies a total length of 1.4–1.5 m (see [34]), is larger than all other fossil species of gars with the exception of the largest known gars in the species †Atractosteus atrox from the Eocene Green River Formation and all extant gars besides A. spatula and L. osseus [34]. This SL greatly exceeds those of other species of Lepisosteus and Atractosteus, as well as those of cuneatin and masillostein gars [34].

In the phylogenetic analysis of gar interrelationships using parsimony, †A. grandei nested within a polytomy formed by all species of Atractosteus, a relationship supported by low bootstrap values (35). This low level of support is probably due to the lack of information on the postcranial anatomy of A. grandei, which reduces the total number of characters coded for this taxon. Nonetheless, the position of this taxon as a species of Atractosteus is consistent across all most parsimonious trees. Bayesian phylogenetic analysis also found †A. grandei within Atractosteus as one node closer to the crown group than the Cenomanian species †A. falipoui [34].

4. Discussion

The discovery of †Atractosteus grandei directly above the K–Pg boundary provides a wealth of new information about the tempo of ecosystem change in the aftermath of the impact and the early evolution of crown-group gars. First, as the oldest definite member of the gar crown group from the Americas (approx. 5 million years older than Atractosteus sp. from the Paskapoo Formation; [34]), †A. grandei conclusively shows that crown lepisosteids have existed in North American river systems since the start of the Cenozoic. Second, †A. grandei displays the oldest case of large body size (greater than 1 m) in the Lepisosteidae.

The preservation of †A. grandei 18 cm (approx. 2–3 thousand years; see electronic supplementary material, table S1) above the K–Pg boundary in the Williston Basin of North Dakota shows that freshwater ecosystems able to support large, approximately 1.5 m-long apex predators like large gars either persisted across the K–Pg boundary or rapidly rebounded within a few thousand years after the K–Pg mass extinction. Freshwater ecosystems saw dramatically lower levels of extinction during the K–Pg event compared to that of marine ecosystems—10% in freshwater ecosystems [47] compared to 50% in marine ecosystems [48]. More recent studies have used high-resolution chronostratigraphic frameworks to analyse the pace of mammalian diversity [27] and maximum mammalian body mass through the first million years of the Cenozoic [28]. Larger (20 kg) mammals do not appear until approximately 300 kyr after the K–Pg mass extinction in the Denver Basin [28]. †A. grandei, living within thousands of years after the K–Pg, would have coexisted with small-bodied (body size less than 1 kg) mammalian ‘disaster’ faunas (e.g. [27]) and depauperate forests that represented an early wave of ecosystem succession in the early Palaeocene [49–53]. With an estimated total length of 1.4–1.5 m, the new gar is therefore one of the largest freshwater vertebrates to appear on the North American landscape directly after the impact. Further, †A. grandei evinces that some clades of freshwater vertebrates show little evidence of a heavy degree of size reduction consistent with an ecosystem-wide Lilliput effect.

In the thousands to hundreds of thousands of years following the extinction, large predatory gars like †A. grandei would have coexisted with a handful of relatively large, freshwater vertebrates, including the giant soft-shelled turtle Axestemys (approx. 1 m long carapace; [28,54,55]) and large-bodied salamanders [56]. The presence of larger vertebrates from disparate clades in freshwater ecosystems supports the idea that freshwater ecosystems functioned as refugia in the wake of the K–Pg extinction [47]. At the same time, the physiological agility of gars cannot be ignored. For example, extant lepisosteids can breathe air, a trait which allows gars to survive in deoxygenated water (e.g. [34,57]). In addition, extant species of the genus Atractosteus are able to tolerate a wide variety of salinities [58]. Finally, all extant gars are generalist predators (e.g. [34]). This suite of features undoubtedly helped gars persist and flourish as ‘disaster taxa’ in the wake of the destruction caused by the asteroid impact, helping this lineage establish a 65-million-year dynasty in North American river systems.

Data accessibility

All associated data are in the manuscript and electronic supplementary material [59].

Authors' contributions

C.D.B.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing—original draft and writing—review and editing; T.R.L.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, resources, software, supervision, validation, visualization and writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

We received no funding for this study.