Marine subsidization of dinosaurian ecosystems

Marine subsidization (MS) of terrestrial ecosystems, in the form of marine-derived organic matter, energy, and nutrients, is a well-known phenomenon on modern coastlines. However, MS is rarely mentioned or considered in studies of Cretaceous terrestrial coastal ecosystems. Many studies of Cretaceous coastal ecosystems have noted `anomalous' carbon isotope (δ 13 C) values much higher than anticipated from comparisons of the bioapatite of dinosaurs and other vertebrates and their presumed dietary δ 13 C value. Since macroalgae and marine plants use a modified C 3 photosynthetic pathway resulting in a relatively increased δ 13 C value, bioapatite of MS consumers will be more enriched in 13 C relative to C 3 consumers. We used stable carbon isotope analysis of tooth enamel to determine if Cretaceous coastal organisms utilize marine resources by comparing coastal to upland ecosystems. These ecosystems include a variety of vertebrate taxa such as fish, neosuchians (e.g., crocodilians), and dinosaurs. The ecosystem deposits are the following: the Dunvegan Formation of British Columbia and Alberta, Canada; the Wayan Formation and Draney Limestone of Idaho, USA; the Antlers Formation of Oklahoma and Texas, USA; the Woodbine Formation/Group and Lewisville Formation of Texas, USA; and the Mussentuchit Member of the Cedar Mountain Formation, Utah, USA. The δ 13 C values of coastal assemblages are compared to assemblages of the Cloverly Formation of Wyoming and Montana, USA, an inland intercontinental deposit where MS was expected to be minimal based on known stratigraphic interpretations. We find evidence of MS of coastal ecosystems with relatively high δ 13 C value of coastal vertebrate enamel (up to >5.5‰ above expected δ 13 C enamel values) but not in upland environments. This has implications for back-calculations of dietary plant δ 13 C value from dinosaur herbivore enamel as previous estimates did not include marine-derived carbon sources in calculations. Therefore, future estimations of Cretaceous vertebrate isotopic offsets must consider the influx of 13 C-enriched, marine-derived organic matter and hence, MS of deep-time terrestrial ecosystems. The effects of MS within Cretaceous ecosystems must also be acknowledged as potential major ecosystem and trophic drivers.

Modern beaches are frequently inundated with allochthonous inputs of marine materials, nutrients, energy, and organisms that cross the marine-terrestrial boundary. These inputs are subsequently incorporated into terrestrial ecosystems. Inundation of allochthonous resources from the marine realm, referred to as marine subsidization (MS), is a well-known phenomenon on modern coastlines and in modern ecological studies (Polis et al., 1997; Colombini and Chelazzi, 2003; Barrett et al., 2005; Paetzold et al., 2008; Spiller et al., 2010; Mellbrand et al., 2011; Piovia-Scott et al., 2011, 2013). Some taxa prefer to consume marine-derived resources (Conradt, 2000; Wheatley et al., 2012) and in some ecosystems this represents a regular process of a ‘healthy’ ecosystem (e.g., Bilby et al., 1996; Barrett et al., 2005; Darimont et al., 2009; Wipfli and Baxter, 2010; Le Gouvello et al., 2017), particularly in ecosystems where marine subsidization has a stabilizing effect on the receiving ecosystem (Huxel and McCann, 1998). Marine resource use observed in modern organisms can be the result of changing environmental dynamics or stressors such as a disappearance/decrease in their terrestrial resources due to shifting prey and/or resource dynamics, and/or a change in habitat size (Anderson and Polis, 1998; Roth, 2003; Adams et al., 2010; Leighton et al., 2024). For example, coastal spiders and scorpions consume more marine-based foods during times of drought when terrestrial productivity is lower than during wet years (Anderson and Polis, 1998). Another change could be the sudden abundance of a preferred or easily obtained resource, or new access to coastal regions, or newly learned behaviors of a coastal-dwelling species (Jefferies, 2000; Reid et al., 2018; Stander, 2019; Lera et al., 2026). Organisms taking advantage of highly available but temporary or seasonal resources like spawning season for turtles and anadromous fish, or consumption of macroalgae and marine plants during winter or summer months (Adams et al., 2010; Ceacero et al., 2014; Le Gouvello et al., 2017; Tischler et al., 2019) can be subsidized by marine resources through these vectors. Sometimes MS of an ecosystem is channeled through the exploitation of stochastically provided nutrients in an otherwise resource limited terrestrial ecosystem such as on islands or resource-poor coastal regions (Barrett et al., 2005; Spiller et al., 2010; Briggs et al., 2012; Contasti et al., 2012; Piovia-Scott et al., 2013; Stander, 2019; Obrist et al., 2022; Greuel, 2025). Mass extinctions and major climate perturbations may drive increased marine resource usage as ecosystems become more stressed over time. Increased reliance on marine resources may be the effect of ongoing trophic cascades or environmental turnover (Polis et al., 1997). MS can stabilize coastal ecosystems that have relatively increased predator populations compared to inland regions (Polis et al., 1997; Adams et al., 2010; Spiller et al., 2010). MS can also support niche expansion of predators (Darimont et al., 2009; Leighton et al., 2024).

Marine plants and macroalgae, as well as prey items that consume marine plants and macroalgae, can be deposited in coastal regions by storms and wave action. MS occurs when these food sources are subsequently consumed by coastal herbivores and carnivores (Adams et al., 2010; Spiller et al., 2010; Piovia-Scott et al., 2013). Marine macroalgae have been abundant since the Precambrian (e.g., Del Cortona et al., 2020; Hou et al., 2022) and use a modified C 3 photosynthetic pathway that results in a relatively high carbon isotope composition (δ 13 C value) similar to the δ 13 C value of terrestrial C 4 plants (Raven et al., 2002; Velázquez-Ochoa et al., 2022). Similarly, freshwater aquatic plants can have relatively high δ 13 C values within an overall range of values overlapping with those of both C 3 and C 4 plants (Keeley and Sandquist, 1992). δ 13 C value is expressed here in the δ notation as shown in Equation 1.

Carbon isotopes in terrestrial ecosystems trace dietary sources and, in fossilized material, can be used to reconstruct paleodiets (Arens et al., 2000). As such, the organic matter and δ 13 C value of skeletal materials from consumers of marine resources reflect this dietary resource and have δ 13 C enamel values higher than those of typical C 3 -plant ecosystems (Clementz and Koch, 2001; Balasse et al., 2005, 2006; Wheatley et al., 2012; Blanz et al., 2020).

The δ 13 C value of vertebrate skeletal apatite, of both teeth and bone, reflects dietary intake (Koch, 1998; Cerling and Harris, 1999; Passey et al., 2005; Kohn, 2010; Cullen, 2023). While plants with C 4 photosynthetic pathways as well as plants with a Crassulacean acid metabolism (CAM) may have existed during the Cretaceous, they occupied a negligible percentage of typical terrestrial ecosystems (Kohn, 2010). A typical δ 13 C organic value (relative to Vienna Pee Dee Belemnite, VPDB) of bulk C 3 plant tissue from the preindustrial era is –26.5‰ with most modern plants falling between –23 to –34‰ (Bender, 1971; Deines, 1980; Nordt et al., 2016). During the Mesozoic, the average value of preserved plant material was –27.4 ± 2.1‰ VPDB with higher values estimated for the Cretaceous between –26.8 to –21.3‰ VPDB (Nordt et al., 2016).

For most modern herbivorous mammals, tooth enamel carbon isotopic composition shows an offset between diet and enamel (Δ 13 C d-e) in which enamel is 12 to 14‰ (or an average of 13‰) higher than their plant food δ 13 C value (Koch, 1998; Bocherens, 2000; Cullen et al., 2023). The soil organic matter δ 13 C value where these plants live should be approximately equal to the average dietary C 3 plant δ 13 C value (Koch, 1998). Therefore, an herbivorous animal’s enamel δ 13 C value should be roughly equivalent to the dietary C 3 plant food δ 13 C value (or soil organic matter δ 13 C value for dietary plant food estimations) plus 12 to 14‰. The whole body composition of the herbivore animal is around 0 to 1‰ higher than the average dietary C 3 plant δ 13 C value (regarded here as negligible) and, similarly, for carnivorous taxa an additional 0 to 2‰ offset exists per trophic level (Bocherens, 2000; Bocherens and Drucker, 2003; Cullen et al., 2020). The difference from the whole body composition to enamel is around 8-10‰ in mammalian carnivores and 12-14‰ in mammalian herbivores (Bocherens, 2000). Other organisms have known Δ 13 C d-e compositions such as neosuchians who have an Δ 13 C d-e of around 8-12‰, turtles with a Δ 13 C d-e of ~12‰, and birds which have Δ 13 C d-e of 11-16‰ (Cullen, 2023; Cullen et al., 2023).

Previous studies indicate organic materials with a wide range of δ 13 C values existed within dinosaurian ecosystems with unknown sources of high δ 13 C material frequently attributed to one or multiple combined factors, including ‘microenvironments’, increased δ 13 C of atmospheric carbon, taxon-specific physiology leading to larger δ 13 C diet-tissue offsets, habitat niche partitioning, canopy effects, dietary plant specializations, osmotic stress of coastal proximity to marine environments, or more recently, freshwater plant dietary intake (Stanton Thomas and Carlson, 2004; Fricke et al., 2008; Amiot et al., 2010, 2015; Crystal et al., 2019; Cullen et al., 2019; Fastovsky et al., 2020; Boyd et al., 2025; Egan et al., 2026). A significant portion of these deposits are coastal (e.g., Fricke and Pearson, 2008; Fricke et al., 2008; Amiot et al., 2010, 2015; Crystal et al., 2019; Cullen et al., 2019; Egan et al., 2026) but some are interpreted to be ‘upland’ or are primarily terrestrial deposits (e.g., Fricke et al., 2008; Amiot et al., 2015; Fastovsky et al., 2020). Fossil deposits that have relatively higher δ 13 C values of teeth with lithologic evidence of coastal environments have been noted in the literature, however, to our knowledge terrestrial taxa consumption of relatively higher δ 13 C found in marine macroalgae/plants and incorporation into coastal ecosystems has not been considered for these values in Cretaceous ecosystems. Rather, unique physiological functions within the dinosaur corporeal materials themselves (Stanton Thomas and Carlson, 2004; Fricke and Pearson, 2008; Tütken, 2011), have been most frequently invoked to explain the high δ 13 C enamel values. It is only in recent years re-investigations of these hypotheses have been performed (i.e., Cullen et al., 2019, 2020, 2023; Egan et al., 2026), that demonstrate a ubiquitous shift in isotopic baseline in these Cretaceous systems and the presence of higher δ 13 C values in multiple co-occurring vertebrate clades (showing it cannot be a unique feature of dinosaurian physiology). A combination of external environmental and ecological factors, alongside increased terrestrial-aquatic resource exchange, is proposed as an alternate explanation for this observed pattern. Hence, environmental factors, such as ancient plants and microorganisms that produced microenvironments of increased δ 13 C values, i.e., methanogenic soils (Fricke, 2025) or freshwater plant consumption (i.e., Boyd et al., 2025), have been suggested as potential explanations of these signals. However, gradients of high δ 13 C values of dinosaur tooth enamel in coastal deposits and low δ 13 C values of dinosaur tooth enamel from upland environments (Fricke et al., 2008) could be the result of the presence of marine-produced organic matter in the environment. Carbon subsidies from the marine realm and associated δ 13 C values have yet to be confirmed as there are many confounding variables such as a wide range of geographic areas, paleolatitudes, ages of deposition (and, therefore, differing δ 13 C values of CO 2 in the atmosphere at time of deposition), unknown and various states of preservation, and varieties of taxa that complicate comparisons between sites. If dinosaurs and other vertebrates that lived in and were buried in Cretaceous terrestrial coastal environments contain an increased δ 13 C value compared to those of upland terrestrial environments, this could suggest influence of allochthonous carbon.

Analysis of carbon isotope composition of bioapatite has shown aquatic and marine resource influence in coastal-dwelling mammals of the Eocene where high δ 13 C values from bioapatite indicate that early sirenians and cetaceans foraged in estuarine and marine environments (Clementz et al., 2006). Other examples of marine resource influence in deep time ecosystems include several North African Early-Late Cretaceous deposits. As reported by Ibrahim et al. (2020), Hassler et al. (2018), and Benyoucef et al. (2015), an abundance of available aquatic and marine resources may have supported an ‘overabundance’ of large predatory dinosaurs and other aquatic predators (turtles, crocodiliforms, and various fish) relative to the number of herbivorous dinosaurs in the Early-Late Cretaceous ecosystems of the Kem Kem Beds of Morocco, Gadoufaoua of Niger, and the Guir basin in Western Algeria. Hassler et al. (2018) note that, from analysis of calcium isotopes in bone and teeth, niche partitioning occurred within the large terrestrial predators particularly between abelisaurid and carcharodontosaurid theropods, who primarily consumed terrestrial resources, versus spinosaurids, who mainly foraged in aquatic environments. Indeed, globally, dinosaurs and other vertebrate skeletal materials and ichnofossils are frequently preserved in coastal deposits similar to where coastal mammals that consume marine resources are located today (Main, 2005; Johnson et al., 2006; Fricke et al., 2008; Krumenacker, 2010; Cullen et al., 2019; Rozada et al., 2021; Tucker et al., 2022).

Storm events, flooding, and wave action, in the modern, drive marine allochthonous resources to shore, and during the Cretaceous, an enhanced hydrologic cycle and higher sea levels led to stronger storm events and more widespread coastal environments than the modern (Ito et al., 2001; Ufnar et al., 2004, 2008; Suarez et al., 2011; Hay et al., 2019). These events likely moved large amounts of marine derived organic matter onto terrestrial beaches and coastal surfaces, as observed today (Polis et al., 1997; Colombini and Chelazzi, 2003; Mellbrand et al., 2011). Marine organic matter was also in large supply during the Cretaceous as there was very high oceanic primary productivity related to ocean anoxic events (Schlanger and Jenkyns, 1976; Jenkyns, 1980, 2010). In Cretaceous deposits, a combination of strong storms and accessible marine matter sometimes resulted in fossils such as ammonites and marine gastropods that are preserved in terrestrial, coastal tree amber (Yu et al., 2019) demonstrating the influx of marine resources into some Cretaceous terrestrial coastal ecosystems.

As coastal areas are major centers of deposition, with intermixing of terrestrial and marine derived carbon, the δ 13 C value of coastal-dwelling organisms reflect a mixture of these carbon sources (Colombini and Chelazzi, 2003; Schlacher and Connolly, 2009; Mellbrand et al., 2011; Wu et al., 2023). Direct evidence of the consumption of crustaceans in megaherbivorous dinosaur diets has been documented in the Late Cretaceous Kaiparowits Formation in Utah, USA (Chin et al., 2017). Parsimoniously, material with increased δ 13 C value from marine-derived sources through direct consumption by macroorganisms (such as the Kaiparowits megaherbivorous dinosaurs) or incorporation into the food chain via saprophagous invertebrates, would be found in the diets of these coastal dinosaurs and other coastal-dwelling vertebrates; as such, we hypothesize that evidence of MS of coastal ecosystems including dinosaurs, neosuchians, and fish is shown by higher δ 13 C values in Albian-Cenomanian coastal ecosystems along the Western Interior Seaway than in contemporaneous, inland ecosystems, and that examining a gradient of coastal to inland assemblages would show relative differences in the effects of this subsidization.

To that end, we test the carbon isotope composition of various vertebrate skeletal materials to detect the presence or absence of dietary incorporation of 13 C-enriched, marine-derived, organic matter in Cretaceous dinosaurian ecosystems. We have selected six coeval North American Albian-Cenomanian formations; five are coastal with varying degrees of marine influence as well as one known almost strictly inland ecosystem with well described lithologic and geologic contexts. Multiple sites have bulk organic matter measurements throughout the formations and are well dated with multiple sites containing rock of similar and overlapping ages of deposition. These deposits include a variety of vertebrate taxa and locations that span the length of the Western Interior Seaway and ancient Gulf of Mexico coastlines as well as some interior upland ecosystems. These are the Dunvegan Formation of British Columbia and Alberta, Canada; the Cloverly Formation of Wyoming and Montana, USA; the Wayan Formation and Draney Limestone of Idaho, USA (collectively referred to here as the Wayan Formation); the Antlers Formation of Oklahoma and Texas, USA; the Woodbine Formation/Group and Lewisville Formation (collectively referred to here as the Woodbine Formation) of Texas, USA, and the Mussentuchit Member of the Cedar Mountain Formation of Utah, USA (Figure 1). The Dunvegan, Wayan, Antlers, and Woodbine formations as well as the Mussentuchit Member represent coastal deposits (Vitali et al., 2002; Main, 2005; Krumenacker, 2010; Andrzejewski and Tabor, 2020; Tucker et al., 2020, 2022, 2023; Andrzejewski et al., 2022; Plint et al., 2023; Lorente et al., 2025). The Cloverly Formation represents an upland, terrestrial deposit prior to seaway excursion and significant marine influence (DeCelles and Burden, 1992; Elliott et al., 2007; D’Emic et al., 2019; Kalu, 2023; Jackson, 2024; Orchard, 2024). All formations chosen here were deposited during the late Albian-early Cenomanian, a period of relatively stable global climate and, importantly, a relatively stable δ 13 C value of atmospheric carbon dioxide relative to other Cretaceous time periods (Bralower et al., 1999; Joo and Sageman, 2014; Herrle et al., 2015; Gale et al., 2020; Nordt et al., 2024). These localities represent deposits preserved after the cool snap in the Aptian and prior to the Cenomanian-Turonian Thermal Maximum (KTM) (Bralower et al., 1999; Joo and Sageman, 2014; Herrle et al., 2015; Gale et al., 2020; Nordt et al., 2024). We compare the average δ 13 C of bioapatite of a variety of taxa from each formation to assess the presence/absence of MS within each ecosystem. To assess diagenesis (or in this context, alteration in the chemical and physical composition of a material by geologic processes over geologic time), we utilized a comparison of the δ 18 O carbonate and δ 18 O phosphate of the fossil materials to empirically-derived relationships within living tissue samples, as well as analysis of Rare Earth Element (REEs) and trace element concentrations.

The δ 18 O values in vertebrate skeletal bioapatite allows for assessment of diagenetic influence. The δ 18 O value of bioapatite generally reflects the δ 18 O value within the main source of water that the organism ingests (with vital effect offsets) (Kohn and Cerling, 2002; Amiot et al., 2007; Coulson et al., 2008; Fricke et al., 2008; Suarez et al., 2014). The δ 18 O value of bioapatite also broadly corresponds to the air temperature of the location of meteoric precipitation from the atmosphere and, hence, correlates to the general latitudinal position of the location of meteoric precipitation (Gat, 1996; Bowen and Wilkinson, 2002). The empirical relationship between phosphate and carbonate δ 18 O values follows a linear relationship (Bryant et al., 1996; Iacumin et al., 1996; Rey et al., 2023). Deviations from this relationship likely represent differential diagenetic influences on each mineral (Bryant et al., 1996; Iacumin et al., 1996; Rey et al., 2023). The carbonate component is more prone to diagenesis than the phosphate component and, as a result, values that do not follow this linear relationship likely represent the results of diagenetic influences (Iacumin et al., 1996; Cullen et al., 2023; Rey et al., 2023).

Samples of bioapatite from a variety of taxa from sites along the Western Interior Basin were sampled for structural carbonate δ 13 C and δ 18 O, phosphate δ 18 O, REEs, as well as strontium and calcium concentrations. Table 1 lists the formation, taxa, and number of isotopic analyses per taxon. Two time slices were investigated; an Albian time slice and an early Cenomanian time slice. Within each time slice, upland versus coastal ecosystems were evaluated.

For the early Albian, the Cloverly Formation, at a site informally called “Crooked Creek” in Northern Wyoming was sampled (sites V62 and V64 recorded by University of Oklahoma’s Sam Noble Museum of Natural History corresponding to Ostrom Unit V, or the Little Sheep Member) (Cifelli et al., 1998; Nydam and Cifelli, 2002; D’Emic et al., 2019; Kalu, 2023). The “Crooked Creek” site represents a terrestrial upland environment. A temporally comparative site is the “Hartman Fossil Site” in the early Albian Antlers Formation (assemblage provided by the Perot Museum of Nature and Science in Dallas, Texas). The “Hartman Fossil Site” was deposited in a deltaic/strandplain ecosystem and represents a coastal ecosystem (Andrzejewski and Tabor, 2020; Andrzejewski et al., 2022). Historically, the Cloverly and Antlers formations and their taxa have been frequently compared and contrasted due to similarity of age, taxa assemblages, and relative geographic proximity (Brinkman et al., 1998; Nydam and Cifelli, 2002; Oreska et al., 2013; Carrano et al., 2016, 2021; D’Emic et al., 2019; Frederickson et al., 2020).

For the late Albian/early Cenomanian, sites V938 and V920 from the Cloverly Formation also represent an upland ecosystem prior to seaway incursion (DeCelles and Burden, 1992; Elliott et al., 2007; D’Emic et al., 2019). They are located on private land within the borders of the Crow Reservation of Montana, USA (Unit VII of Ostrom (1970) or the Himes Member which overlies the Little Sheep Member) (Ostrom, 1970; D’Emic et al., 2019). They are contrasted with the coastal early Cenomanian Woodbine Formation at the “Arlington Archosaur Site” whose assemblage was provided by the Perot Museum of Nature and Science (Scherer, 2014; Andrzejewski and Tabor, 2020; Noto et al., 2022; Lorente et al., 2025). Additionally, analyses of bioapatite from the Dunvegan Formation of British Columbia and Alberta, Canada were also completed (fossils provided by the Philip J. Currie Dinosaur Museum of the University of Alberta and the Royal Tyrrell Museum). The Dunvegan Formation represents a northern, coastal locality with frequent marine incursions (McCarthy et al., 1999; Vitali et al., 2002; Cook et al., 2008; Ufnar et al., 2008). Analysis of two other low-lying coastal deposits relatively near to the Cloverly Formation were also conducted. These are the Wayan Formation and Draney Limestone of Idaho, USA (Krumenacker, 2010; Ross et al., 2017) (the fossils of which are provided by the Idaho Museum of Natural History within the Idaho State University system) and the Mussentuchit Member of the Cedar Mountain Formation (CMF) of Utah, USA [fossils were provided by North Carolina Museum of Natural Sciences and University of Oklahoma’s Sam Noble Museum of Natural History (Cifelli et al., 1999; Tucker et al., 2020, 2022, 2023)]. Importantly, none of the sites analyzed were deposited simultaneously with the Cretaceous Thermal Maximum at the Cenomanian-Turonian boundary. These six formations represent a variety of mid-Cretaceous sites with overlapping ages and taxa compositions and cover a range of paleolatitudes.

δ 13 C and δ 18 O values of the carbonate fraction

The calcium carbonate component of the vertebrate materials were analyzed for δ 13 C values and δ 18 O values for each of the formations. A total of ~3–5 mg of powder was generated by using a diamond tipped dental drill from each specimen (if the sample was a tooth, dentin and enamel were sampled separately where possible). An aliquot of 0.5 mg was separated for phosphate δ 18 O analysis (detailed below). For analysis of the carbonate molecule, samples were prepared based on methods from Koch (1998). Samples were cleaned of diagenetic carbonates using 1M calcium acetate-acetic acid buffer solution at 0.04 mL/mg. Samples were then reacted for 12 hours, rinsed five times in deionized water, and dried at 49 °C in a laboratory oven. Samples were then screened for any exogenous material, re-crushed, and weighed into 4.5 mL Labco Exetainer containers. Samples were then either reacted with 100% phosphoric acid at room temperature for 24 hours and analyzed on a Gas Bench II coupled to a ThermoFinnigan Delta Plus XP IRMS housed at the University of Arkansas Stable Isotope Lab (UASIL), or reacted with 102% orthophosphoric acid at 70°C for 20 minutes in an online NuCarb automated system coupled to a Nu Perspective IS IRMS housed at the North Carolina State University Paleo 3 Lab. Calibration standards included NBS 19 (δ 13 C=+1.95‰ VPDB, δ 18 O=28.65‰ VSMOW) and in-house standards UASIL 22 (δ 13 C=–35.60‰ VPDB, δ 18 O=–17.07‰ VPDB), UASIL 23 (δ 13 C=–0.60‰ VPDB, δ 18 O=–14.71‰ VPDB) (Yamamura and Suarez, 2025) and Paleo 3 C64 (δ 13 C=–2.11‰ VPDB, δ 18 O=–15.63‰ VPDB). Precision was monitored via analysis of NIST 120c Florida phosphate rock. Average values of these standards were: NBS 19 (δ 13 C=+1.96 ± 0.07‰ VPDB, δ 18 O=+28.66 ± 0.14‰ VSMOW), UASIL 22 (δ 13 C=–35.6 ± 0.08‰ VPDB, δ 18 O=+13.3 ± 0.05‰ VSMOW), UASIL 23 (δ 13 C=–0.61 ± 0.07‰ VPDB, δ 18 O=+15.73 ± 0.12‰ VSMOW), and NIST 120c (δ 13 C=–6.35 ± 0.07‰ VPDB, δ 18 O=+29.5 ± 0.19‰ VSMOW).

δ 18 O values of phosphate

Samples of the phosphate component were prepared based on methods from Vennemann et al. (2002) and Bassett et al. (2007). Approximately 0.3 to 0.5 mg of bioapatite powder was dissolved in 0.5 M HNO 3. Once dissolved, calcium was extracted using an exchange reaction with 0.5 M KOH and 0.17 M KF solutions. Samples were centrifuged and the supernatant was extracted and placed into new 1.7 mL vials. Ag 3 PO 4 was precipitated by addition of a silver amine solution to the supernatant and evaporated in an oven at 49 °C for ~20 hours to reduce pH, allowing Ag 3 PO 4 to precipitate. Crystals were rinsed five times and dried in a laboratory oven before being weighed into silver capsules for analysis.

Stable isotopic analysis was conducted at UASIL and at the stable isotope laboratory at Boise State University. Phosphate δ 18 O values was analyzed at 1400 °C on a high-temperature conversion elemental analyzer (TC/EA) attached to a Thermo Delta Advantage Plus isotope ratio mass spectrometer located at UASIL or on a TC/EA attached to a ThermoFisher Delta V Plus continuous flow isotope ratio mass spectrometer located at Boise State University. All samples are reported relative to the VSMOW. Calibration standards included USGS 80 (+13.1‰ VSMOW), USGS 81 (+35.4‰ VSMOW), NBS-127 barium sulfate (+9.3‰ VSMOW) and NIST 120c (+21.7‰ VSMOW), and NIST 120c was also analyzed with each batch of samples as a procedural standard. Average δ 18 O values of these standards were +13.4 ± 0.57‰ VSMOW, + 35.0 ± 0.57‰ VSMOW, + 7.65 ± 0.41‰ VSMOW, and +22.4 ± 0.57‰ VSMOW respectively.

Comparison of carbonate versus phosphate δ 18 O (Figure 2) was used to assess diagenetic alteration (Cullen et al., 2019, 2023). A difference between minerals following Equation 2 from Iacumin et al. (1996),

should be observed in unaltered material (Bryant et al., 1996; Iacumin et al., 1996; Rey et al., 2023). The linear relationship between mammalian enamel and bone δ 18 O carbonate vs δ 18 O phosphate defined by Iacumin et al. (1996) is used here as a baseline to indicate effects of diagenesis. While no mammalian fossils were analyzed in this study, the mammalian enamel and bone δ 18 O carbonate vs δ 18 O phosphate relationship is still a useful comparison as dinosaur and reptile δ 18 O bioapatite values approximately parallels the mammalian δ 18 O relationship (Amiot et al., 2010). The linear relationship defined by Iacumin et al. (1996) has also been referenced in numerous dinosaur and other deep-time vertebrate studies (e.g., Lécuyer et al., 2003; Amiot et al., 2010; Rey et al., 2023) so its inclusion also allows for comparisons between these studies. Differences between phosphate-carbonate should be approximately ~3-10‰ depending on the temperature of formation and species specific diet and physiology (Bryant et al., 1996; Iacumin et al., 1996; Zazzo et al., 2004; Göhring et al., 2019).

Laser ablation analyses

REEs (lanthanum, ytterbium, samarium), strontium, and calcium concentrations of select specimens of bioapatite were analyzed via laser ablation system (ESI NWR193 nM excimer laser with these settings: spot size 50μm, laser rep rate of 10 Hz, burst of 200 shots per 20 seconds, fluence at 40% ~4.5J/cm^2, a typical analysis of 15 seconds of gas blank, followed by 20 seconds laser on, finally followed by 10–15 seconds of washout) attached to Thermo iCapRQ quadrupole inductively coupled plasma mass spectrometer (ICP-MS) at the Trace Element and Radiogenic Isotope lab (TRAIL) at the University of Arkansas. Trace element glass standards NIST 610 and 612 were used as calibration standards and the internal standard was Ca at 38 weight percent. Samples were mounted on adhesive putty, photographed, and submitted for analysis. Following a similar method to Cullen and Cousens (2024), analysis of La/Yb and La/Sm ratios normalized to North American Shale Composite (NASC) were conducted to also assess preservation status and the impacts of REE adsorption, substitution processes and late stage diagenesis where available (Picard et al., 2002; Reynard and Balter, 2014; Cullen and Cousens, 2024).

The use of Sr/Ca ratios observed in the composition of fish scales and otoliths can be used to determine the geographic source of a fish (Courtemanche et al., 2005; Macdonald and Crook, 2010; Seeley et al., 2015, 2017; Daugherty et al., 2017, 2019). This is accomplished by analysis of the Sr/Ca ratios which is positively correlated to salinity, higher salinity = higher Sr/Ca ratio. Sr/Ca ratios of ~1-1.5 indicate low salinity (0-5‰) and higher salinity (>5‰) correlates to Sr/Ca ratios of >2mmol/mol. Sr/Ca values in between 1.5–2 mmol/mol indicate marginal and brackish conditions (Macdonald and Crook, 2010).

Isotopic compositions

δ 13 C and δ 18 O value averages and standard deviations per formation and by specimen are shown in Table 2, Supplementary Table 1, and Supplementary Table 2. The δ 13 C values from all formations range from –19.7 to +0.63‰ VPDB. The δ 18 O carbonate values of all formations range from +15.86 to +34.35‰ VSMOW. The range of δ 18 O phosphate values of all formations is +8.7 to +24.3‰ VSMOW. Cross-plots of δ 18 O carbonate vs δ 18 O phosphate are shown in Figure 2. Figure 2A includes all analyses of bone, dentin and enamel compared to the empirical relationship of δ 18 O carbonate to δ 18 O phosphate derived by Iacumin et al. (1996). Figure 2B shows only the δ 18 O values of enamel from these specimens. The spread is much tighter within the enamel-only plot compared to all bioapatite components. Nearly all enamel values fall along the empirical relationship derived by Iacumin et al. (1996) whereas some bone and dentin values fall to the right and below the linear relationship. The Dunvegan Formation assemblage is removed from the enamel-only plot as it is only composed of bone specimens. Figure 3 presents δ 13 C carbonate relative to δ 18 O phosphate for all samples (3A) and enamel only samples (3B). The ranges of all assemblages contract with the removal of bone and dentin from the plots but the general relative relationship to other formations remains nearly the same. The mean δ 13 C value of the enamel assemblages from the coastal formations of the Woodbine, Antlers, and Wayan, are more positive than the average value of the upland Cloverly Formation assemblage during both the early Albian and the late Albian.

A comparison of all samples, including enamel, dentin, and bone for all sites were statistically evaluated to allow for greater sample sizes and less likelihood of type I error. Utilizing Welch’s (Two Sample) t-test with a p-value threshold of 0.05, there is no statistically significant difference between the Little Sheep Mudstone and Himes Member of the Cloverly Formation in δ 13 C values of all analyses (p-value= 0.177, average δ 13 C of the Himes= –9.92‰ VPDB, average δ 13 C of the Little Sheep Mudstone = –9.27‰VPDB). There is statistical significance in differences between the average Cloverly Formation δ 13 C value (–9.36‰VPDB) and the following formations: the Antlers Formation (–5.54‰ VPDB, p-value: <0.0001); the Wayan Formation (–6.53‰ VPDB, p-value: <0.0001); the Woodbine Formation (–4.60‰ VPDB, p-value: <0.0001); the Dunvegan Formation (–3.02‰ VPDB, p-value: <0.0001); and the Mussentuchit Member of the CMF (–6.21‰ VPDB, p-value: <0.0001).

Using a p-value threshold of 0.05, the coastal deposits of the Antlers and Woodbine formations are not statistically different in δ 13 C value range (p-value: 0.054) and neither are the Antlers Formations and the Mussentuchit Member (p-value: 0.070). If using a p-value threshold of 0.01, the coastal Woodbine and Dunvegan formations are not statistically significantly different (p-value: 0.041) and the Wayan and Antlers formations are not statistically significantly different (p-value: 0.040). At p-value threshold of 0.01, all coastal formation assemblages have mean values that are not dissimilar but all are statistically different from the Cloverly Formation assemblage mean. F test results (conducted with the var.test function from the ‘stats’ package in R (The R Core Team, 2025)) are also included in Supplementary Table 2. All but the Wayan-Cloverly (p-value: 0.0326) and the Wayan-Woodbine (p-value: 0.0244) comparisons were not statistically significant in δ 13 C values.

The averages of the Wayan and Dunvegan formations in δ 18 O phosphate space are not statistically significantly different (δ 18 O phosphate = +10.79‰ VSMOW, δ 18 O phosphate = +10.54‰ VSMOW, respectively, with a p-value: 0.690) as well as the Woodbine Formation and Mussentuchit Member (δ 18 O phosphate =17.67‰ VSMOW, δ 18 O phosphate =17.54‰ VSMOW, p-value: 0.664) All other formation δ 18 O phosphate mean combinations were significantly different as can be seen in Figure 2 and Supplementary Table 2.

Elemental concentrations & ratios

The specimen OMNH 64911, is a single specimen of Lepisosteidae scale from Ostrom Unit VII or the upper Himes Member of the Cloverly Formation (Ostrom, 1970; D’Emic et al., 2019). It is larger with a maximum dimension of 19 mm across. From 20 laser ablation analyses, the average Sr/Ca ratio of OMNH 64911 in mmol/mol is ~2.85 ± 0.19 (minimum=2.56, maximum=3.45) (Supplementary Table 3). The same analyses were conducted on two Lepidotes (?) teeth from the Antlers Formation, PMNS 2007-11-070 (formerly DMNH 2007-11-070), which had average Sr/Ca ratios of ~5.98 and ~5.02 mmol/mol from 3 laser ablation analyses each.

Shown in Figure 4, the (La/Sm) N and (La/Yb) N ratios of the Cloverly Formation assemblages plot between modern river and lake as well as modern estuarine/coastal. The Antlers Formation assemblage is spread across multiple envelopes. It plots within Cenozoic marine, Quaternary marine, modern estuarine/coastal, and modern river and lake. Bone and dentin from the Antlers Formation assemblages have two centers of concentration within the Cenozoic marine and Quaternary marine. These do not appear within the enamel only plot. The Woodbine Formation assemblage almost exclusively plots within modern river and lake as well as Quaternary marine envelopes. Some bone and dentin specimens have concentrations trending towards a metamorphic (recrystallization) trend but also mostly disappear within the enamel only plot. The Wayan Formation assemblage bone and dentin plot distinctly within the Mesozoic marine envelope and also trend in the direction of inorganic adsorption effects. The Wayan Formation enamel-only specimens have a few analyses that plot in the Mesozoic marine envelope as well as Cenozoic marine, and modern estuarine/coastal. The Dunvegan Formation assemblage, which is entirely bone, plots within Mesozoic marine, modern river and lake, and modern estuarine/coastal envelopes.

Diagenesis

To screen out any samples that might have diagenetic alteration, comparisons of (La/Sm) N and (La/Yb) N ratios (normalized to North American Shale Composite) as well as comparison of δ 18 O values of phosphate and carbonate were conducted (Figure 2). A 1-3‰ offset in both δ 13 C and δ 18 O values is typically found between unaltered enamel and dentin (Owocki et al., 2020; Cullen et al., 2023). Dentin is more likely to be diagenetically altered (Owocki et al., 2020; Cullen et al., 2023). Differences in the stable isotope ratios between enamel and dentin within tooth bioapatite should be expected whereas homogeneity of stable isotope ratios, as well as within each mineral sample, may be the result of diagenetic overprinting (Sharp and Cerling, 1998; Cullen et al., 2023). As can be seen in Figure 3 and Table 2, dentin and bone values, on average, are higher in δ 13 C value by ~1-3‰ than enamel compositions and when removed from the calculation of the average δ 13 C value, the average is lowered. This is not unusual and likely indicates the majority of the samples have not undergone significant diagenesis. Enamel has fared the best over time with most samples being unaltered. A possible light alteration in the dentin fraction may be present (Owocki et al., 2020; Cullen et al., 2023). In addition, (La/Sm) N and (La/Yb) N cross-plots have been frequently used to determine diagenetic status (Picard et al., 2002; Reynard and Balter, 2014; Martin et al., 2022; Cullen and Cousens, 2024).

REE concentration is not necessarily a perfect determination of diagenesis. Indeed, the isotopic composition of the bioapatite crystals may retain the same isotopic composition even with the introduction of REE-rich fluids (Tütken (2011) and references therein). However, the concentrations in dentin and bone especially can reveal their burial history and some diagenetic influences (Tütken, 2011). A review of the (La/Sm) N and (La/Yb) N ratios indicates moderate recrystallization in some specimens of dentin and bone in the Woodbine Formation assemblage and moderate adsorption in a few specimens in the assemblage from the Wayan Formation with a significant influence of Mesozoic marine conditions (Reynard and Balter, 2014; Martin et al., 2022; Cullen and Cousens, 2024) (Figure 4). A vast majority of the data overall do not suggest significant amounts of diagenesis within (La/Sm) N and (La/Yb) N ratio space as most plot within a marine, river, or lake environment as expected for their burial locations (Cullen and Cousens, 2024). The Cloverly Formation assemblage appears to have an overwhelming signal of terrestrial environments with some potential of estuary/coastal influence (Martin et al., 2022; Cullen and Cousens, 2024). The isotopic composition of enamel from the Antlers Formation assemblage shows the influence of marine subsidization which is consistent with its inferred depositional environment. The same is true for the Woodbine Formation enamel-only analyses, although, some recrystallization within the more porous specimens is possible. The Wayan Formation assemblage is split between marine and estuarine/coastal influences which agrees with its inferred depositional environment. No significant deviations from inferred depositional environments and (La/Sm) N and (La/Yb) N are observed here.

The Mussentuchit Member assemblage was not analyzed via laser ablation as the original specimens were not available but all assemblages were analyzed for δ 18 O carbonate and δ 18 O phosphate values. Most samples from all sites do not appear to be significantly altered in δ 18 O carbonate and δ 18 O phosphate space. This was determined by the comparison of samples to the Iacumin et al. (1996) empirical relationship between δ 18 O phosphate and δ 18 O carbonate. δ 18 O phosphate and δ 18 O carbonate in bioapatite materials of extant mammals are positively correlated as both minerals of the teeth are precipitated close to equilibrium with the organism’s body water so deviations from that relationship would suggest diagenetic influence (Iacumin et al., 1996; Zazzo et al., 2004; Rey et al., 2023). As δ 18 O carbonate is more prone to diagenetic influences, samples that fall significantly to the right of the Iacumin et al. (1996) empirical relationship in Figure 2 likely indicate diagenetic influences possibly due to microbially-induced isotopic exchange (Bryant et al., 1996; Iacumin et al., 1996; Zazzo et al., 2004; Rey et al., 2023). As dentin/bone have been shown to be less robust over geologic time, and since removing them eliminates the vast majority of samples to the right of the line, only enamel δ 13 C carbonate and δ 18 O phosphate values will be used for our interpretations henceforth (Trueman and Tuross, 2002; Zazzo et al., 2004; Fricke et al., 2008; Dodat et al., 2023; Rey et al., 2023).

Marine subsidies

The Δ 13 C d-e for hadrosaur dinosaurs was originally hypothesized as the difference between the bulk sedimentary organic matter and the tooth enamel of hadrosaur dinosaurs preserved together (Stanton Thomas and Carlson, 2004; Fricke and Pearson, 2008), suggesting a possible Δ 13 C d-e of 18‰ VPDB. Later analysis of the dietary preference of sauropod dinosaurs compared carbonized fossil leaves and sauropod bioapatite (Tütken, 2011) estimated a Δ 13 C d-e of 16‰ VPDB. Later assessments of the δ 13 C values within a microfossil assemblage in the Oldman Formation of Alberta, Canada, Cullen et al. (2023) identified an ecosystem-wide anomalous increase in δ 13 C values across all taxa. A decoupling of the δ 13 C values of in-situ bulk organic matter and the composition of various organismal materials has been postulated to be one cause of relatively high δ 13 C values in enamel composition (Tütken, 2011; Cullen et al., 2023). A decoupling like this in modern ecosystems is a hallmark of incorporation of 13 C-enriched marine carbon and hence, subsidization by marine resources (Adams et al., 2010; Angerbjörn et al., 2024).

For example, the fur of a fox population in the mountains of Sweden is significantly higher in δ 13 C value than is typical for a non-subsidized fox population in a C 3 -dominated ecosystem (Angerbjörn et al., 2024). In this case, its dietary preference of migratory seabirds, which have higher δ 13 C values relative to the average local fox prey, contributes to a decoupling of the fox δ 13 C value and the local δ 13 C value average (Angerbjörn et al., 2024). Dietary preferences and geographic proximity to marine environments may control the type and degree of MS (Korobushkin et al., 2016; Schindler and Smits, 2017; Montagano et al., 2019; Obrist et al., 2022), but marine resource consumption is seen in numerous groups of large and small herbivorous animals over modern and archaeological time periods such as in cervids and relatives, ungulates, avians, and potentially mammoths (Hall and Moore, 1986; Conradt, 2000; Balasse et al., 2005; Lysak, 2013; Ceacero et al., 2014; Bakker et al., 2016; Tischler et al., 2019; Blanz et al., 2020; Takahashi et al., 2024; Greuel, 2025). Marine resource consumption in these organisms results in elevated δ 13 C values relative to their baseline C 3 -dominated ecosystem counterparts. Omnivorous and carnivorous terrestrial vertebrates frequently consume marine-influenced resources as well. These include a variety of taxa such as carnivorans, neosuchians, various lizards and other small reptiles, rodents, and turtles (Hilderbrand et al., 1996; Colombini and Chelazzi, 2003; Stapp and Polis, 2003; Barrett et al., 2005; Darimont et al., 2009; Schlacher and Connolly, 2009; Spiller et al., 2010; Piovia-Scott et al., 2011, 2013; Wheatley et al., 2012; Le Gouvello et al., 2017; Richardson et al., 2019; Stander, 2019; Davidson et al., 2021; Angerbjörn et al., 2024; Leighton et al., 2024; Gerraty et al., 2025). Various producers, and primary and secondary consumer invertebrates are also influenced by marine resources (Colombini and Chelazzi, 2003; Paetzold et al., 2008; Mellbrand and Hambäck, 2010; Spiller et al., 2010; Piovia-Scott et al., 2011). Populations that are subsidized by allochthonous resources may be up to 57% larger due to the subsidization (Montagano et al., 2019; Greuel, 2025).

As mentioned above, Cretaceous dinosaurian ecosystems appear to have a decoupling of bulk organic matter and vertebrate bioapatite with numerous examples of anomalous average δ 13 C values (Fricke et al., 2008; Amiot et al., 2015; Crystal et al., 2019; Fastovsky et al., 2020; Cullen et al., 2023). Fricke et al. (2008) note that the closer to the coastline a deposit is located, the higher the average δ 13 C in herbivorous taxa. Egan et al. (2026) suggest coastal proximity played a major role in the isotopic compositions and elemental concentrations within the Campanian Dinosaur Park Formation taxa of Saskatchewan, Canada. A decoupling of δ 13 C in herbivorous taxa and bulk organic carbon in soils and atmosphere is also seen in the coastal ecosystems of the Antlers, Woodbine, Wayan, Dunvegan, and Mussentuchit formations/member ecosystems (Figure 5).

Comparison between the average δ 13 C value of bulk organic matter of the formations and the average δ 13 C of herbivorous dinosaur vertebrate enamel is shown in Figure 6. This is the same comparison conducted by Fricke and Pearson (2008) that established the hypothesized Δ 13 C d-e relationship for hadrosaur tooth enamel to bulk organic matter. The average herbivorous dinosaurian enamel of the Antlers Formation is –8.28 ± 1.86‰ VPDB, the Cloverly Formation is –10.27 ± 0.99‰ VPDB, the Woodbine Formation is –3.58 ± 1.04‰ VPDB, and the Mussentuchit Member is –6.35 ± 1.84‰ VPDB. An average baseline of δ 13 C in C 3 plants ranges between –26 to –27‰ VPDB today (Tütken, 2011; Nordt et al., 2016) but the average bulk organic matter in Cretaceous δ 13 C chemostratigraphic records are much more positive. The bulk organic matter δ 13 C average within the Cloverly Formation was –24.1 ± 1.37‰ as seen in sites V62 and V64 (“Crooked Creek”) (Kalu, 2023). Approximately 1,100 km to the south of the Antlers Formation sites, chemostratigraphic columns from Santa Rosa Canyon in northern Mexico observed average δ 13 C value of bulk organic matter ~–23‰ VPDB or about +3-4‰ difference as compared to today’s current average (Bralower et al., 1999). Similarly, the bulk organic matter of the Mussentuchit Member has an average δ 13 C value of –24.04 ± 1.53‰ VPDB (Forster, 2022). Using a modern C 3 herbivore Δ 13 C d-e of 13‰, the calculated δ 13 C enamel for an ecosystem with an average of –24‰ VPDB would be ~–11‰ VPDB, consistent with the Cloverly Formation herbivore average but is much lower than the Antlers Formation herbivore average (Figure 6). If a Δ 13 C d-e of hadrosaurs of +18‰, is used, the expected enamel value should be ~–6‰ which is higher than Cloverly Formation (δ 13 C herbivore =–10.27 ± 0.99‰ VPDB) and Antlers Formation (δ 13 C herbivore =–8.28 ± 1.86‰ VPDB) herbivore averages but is within one standard deviation of the Mussentuchit Member average (δ 13 C herbivore =–6.35 ± 1.84‰ VPDB) (Figure 6). An 18‰ Δ 13 C d-e for herbivorous dinosaurs appears to approximate the δ 13 C enamel of coastal herbivores such as the Antlers Formation and Mussentuchit Member, however, it significantly overestimates the average in upland environments such as in the Cloverly Formation assemblage. A similar situation is noted in analysis of hadrosaur δ 13 C enamel from Baja California, Mexico by Fastovsky et al. (2020).

Fastovsky et al. (2020) excavated hadrosaurine teeth from El Gallo Formation, Baja California, Mexico and noted a discrepancy from the expected geologic interpretation and the δ 13 C enamel of the hadrosaur teeth. A common explanation for a lower than expected δ 13 C value of, in this case, hadrosaur enamel is due to carbon recycling in the soil resulting in a “canopy effect” (Cerling et al., 2004). A canopy effect occurs when a significant difference in the δ 13 C value of leaves exists at the tops of canopies versus near the forest floor. The δ 13 C leaf value is lower in the subcanopy environments where recycling of soil-respired carbon dioxide occurs resulting in very low δ 13 C value of soil organic matter. The interpretation of the local geology where the hadrosaur enamel from Baja California was excavated is incongruent with such an interpretation. The implied depositional system is braided streams and fluvial systems with high flow intensities which are not conducive to recycling of soil-respired carbon dioxide needed for a canopy effect. The average of six hadrosaur tooth enamel from the El Gallo Formation δ 13 C values is –9.33‰ VPDB (Fastovsky et al., 2020). Subtracting a mammalian Δ 13 C d-e of 13-14‰ would suggest a bulk organic matter δ 13 C value of –22 to –24‰ VPDB. This is similar to the difference between the Cloverly Formation herbivore average δ 13 C enamel value and the bulk organic matter δ 13 C value and is also more aligned with the geologic interpretation of the El Gallo Formation. This suggests little to no MS of the El Gallo Formation ecosystem is implied by the El Gallo Formation hadrosaurs. A possible explanation for the Δ 13 C d-e overestimation of hadrosaurine enamel could be that the +18‰ Δ 13 C d-e offset the authors used was derived from hadrosaur enamel grown after consuming marine carbon or freshwater plants. A hidden offset of around +4-5‰ from the consumption of 13 C-enriched marine organic matter, and/or an overall preference for freshwater macroalgae/plants with higher δ 13 C value (Schoeninger and DeNiro, 1984; Boyd et al., 2025) could explain the difference between modern mammalian Δ 13 C d-e of 13-14‰ and the +18‰ Δ 13 C d-e offset previously suggested. This is particularly relevant as the Hell Creek Formation, where the hadrosaur Δ 13 C d-e was proposed, is a well-known coastal ecosystem (Hoganson and Erickson, 2005; Fricke and Pearson, 2008; Fowler, 2020). This discordance is very clear in the early Cenomanian Woodbine Formation where comparisons of charcoal and herbivorous dinosaurs are the most discrepant.

The Woodbine Formation assemblage is constrained to a site in Arlington, Texas, known as the Arlington Archosaur Site (AAS) (Lorente et al., 2025). δ 13 C values from charcoal within the rock at the AAS contain an average value of –24.88 ± 0.67‰ (Scherer, 2014). Charcoalified plant matter is typically within 1‰ of the original plant material with a possible loss of 12 C resulting in remnants 1‰ higher in δ 13 C value (Nordt et al., 2016). This suggests the original plant matter could be as low as –26.55‰ VPDB. A Δ 13 C d-e of +18‰ suggests hadrosaur δ 13 C enamel should be ~–7‰ VPDB. However, the average measured hadrosaur δ 13 C enamel is –4.26 ± 0.55‰ VPDB or an approximate +3‰ difference from the expected value, and for a Protohadros sp. (–2.17‰ VPDB) specimen, an approximate +5‰ difference suggests a secondary source of organic carbon was consumed. These average δ 13 C enamel values are comparable to marine sirenians (almost exclusively consuming marine macroalgae and plants) of the Eocene (0 to –3‰VPDB) (Clementz et al., 2006).

Marine carbon isotopic ratios are inherited from the base of the food-chain in primary producers to herbivores to carnivores (Paetzold et al., 2008; Spiller et al., 2010). The difference between predator-prey trophic level enrichment is 0 to +2‰ so the difference between δ 13 C enamel and δ 13 C bulk organic matter should be approximately the same for carnivores as it is for herbivores with a potential slight increase of up to 2‰ from herbivore to carnivore (Bocherens and Drucker, 2003). Using the highest δ 13 C organic matter from the AAS of ~–24‰ VPDB as well as the maximum crocodilian Δ 13 C d-e of ~11.5‰ (Cullen, 2023), the expected δ 13 C enamel for the Woodbine Formation crocodiles should be –12.5‰ VPDB, a 7.5‰ difference from the measured crocodilian enamel average ~–5‰ VPDB. This same pattern can be seen in an inland vs coastal crocodile/alligator comparison today in Florida where coastal alligators are 5.4‰ higher than their inland counterparts (with an estimated 65% dietary intake consisting of marine-based prey items) in δ 13 C value in bioapatite in teeth (Wheatley et al., 2012; Cullen, 2023). The same interpretation holds true for the Wayan Formation, Antlers Formation, and Mussentuchit Member crocodilians, albeit at smaller magnitudes than the Woodbine Formation crocodiles (Figure 5, Figure 6). The difference from the average δ 13 C enamel (–10.5 ± 0.12‰) and expected δ 13 C enamel (~–12.6‰) of the Cloverly Formation crocodilians is ~2‰ difference which is the smallest offset from average to predicted values and within the known bulk organic matter to predator Δ 13 C d-e.

An alternative hypothesis for the greater than expected average δ 13 C enamel values of Early Cretaceous ecosystems investigated here, is that the average δ 13 C atmosphere value during the Cretaceous was higher than it is today. The atmospheric δ 13 C value can be estimated using δ 13 C values of Tethyan marine bulk carbonates as the atmospheric δ 13 C value is reflected in the dissolved inorganic carbon and subsequently recorded in the marine bulk carbonates (Barral et al., 2017). Indeed, the δ 13 C atmosphere during Ocean Anoxic Event 2 may have been as high as –2.57‰ VPDB during the early Turonian (Barral et al., 2017). However, all of the ecosystems investigated here were deposited during a range of δ 13 C atmosphere values between –5.97 and –4.99‰ VPDB in the early Albian to early Cenomanian (Barral et al., 2017) and an atmosphere more enriched in 13 C would not explain the discrepancy between upland versus coastal ecosystems that were deposited at the same time. An enriched δ 13 C atm would have enriched both ecosystems. Shifting atmospheric δ 13 C values also does not explain the similarity of higher δ 13 C values of enamel samples from the Albian-Cenomanian ecosystems and later Campanian-Maastrichtian ecosystems, where lower atmospheric δ 13 C values were prominent (Barral et al., 2017) but where anomalous δ 13 C enamel values are also observed (Cullen et al., 2023).

While specific groups of taxa such as neosuchians may simply prefer to consume a diet composed primarily of marine resources (e.g., modern American crocodiles with δ 13 C bioapatite = –9.6 ± 3.9‰ VPDB), or a diet of mixed terrestrial-marine resources similar to the diets of modern American alligators (mixed~ –10.3 ± 2.6‰ VPDB vs inland~ –15.7 ± 1.8‰ VPDB) (Wheatley et al., 2012), this may not apply to all taxa. Coastal taxa may not prefer or are not able to consume marine macroalgae/plant or other marine-derived material and, hence, would not exhibit increased δ 13 C bioapatite compared to their counterparts consuming marine resources. The existence of these taxa within these assemblages would support the hypothesis of marine resource consumption over a strictly environmental signal such as a relatively increased atmospheric δ 13 C value.

Isotopically irregular taxon-specific interpretations

Tenontosaurus tilletti

One taxon that does not appear to consume marine organic matter in the composition of their δ 13 C enamel value but is nonetheless included in the “coastal” taxa, is Tenontosaurus tilletti. This specimen is noteworthy and is key to the MS hypothesis. Increased δ 13 C enamel value of taxa is not necessarily omnipresent in every coastal organism. Likewise, not every aquatic plant has a relatively increased δ 13 C value (Keeley and Sandquist, 1992; Boon and Bunn, 1994; Finlay and Kendall, 2007) (Figure 6). For those taxa that prefer these aquatic plants, they should reflect the lower δ 13 C value in their skeletal material and collagen (e.g., Schoeninger and DeNiro, 1984; Cullen et al., 2023). Indeed, some aquatic plants can have δ 13 C organic matter values that fall within the range of terrestrial C 3 plant compositions (Rounick and Winterbourn, 1986; Keeley and Sandquist, 1992; Boon and Bunn, 1994; Finlay and Kendall, 2007). Hippopotamus antiquus, a widely-regarded aquatic herbivore, of lower Pliestocene Spain, had a δ 13 C collagen value indistinguishable from the δ 13 C collagen value of strictly terrestrial taxa despite its aquatic plant dietary preference not seen in other coeval herbivores (Palmqvist et al., 2003). Dietary preference is a key part of the equation and if a preferred food source is abundant then consumption of marine macroalgae and plants may not be necessary or preferred and, likewise, not observed. For this reason, the geologic and geographic interpretations are necessary to determine the likely source of dietary organic matter.

In the coastal Antlers Formation assemblage, enamel of a T. tilletti tooth specimen has the lowest δ 13 C value of –9.6‰ with a similar enamel δ 13 C value of T. tilletti from the inland Cloverly Formation assemblage at –10.5 ± 0.56‰. Using a Δ 13 C d-e of –13‰ for a typical C 3 plant consuming mammal (Koch, 1998), the Antlers Formation δ 13 C plant average is calculated to be ~–21 to –23‰ similar to the bulk organic matter of the chemostratigraphic section of the Cloverly Formation, and for the corresponding time period observed in the Santa Rosa Canyon chemostratigraphic column (Bralower et al., 1999). Similarly, within the Cloverly Formation assemblage, T. tilletti averages of δ 13 C enamel is –10.5 ± 0.56‰ which implies a diet of C 3 plants with an average δ 13 C value of –23.5‰ VPDB and which is within one standard deviation of the average bulk organic matter δ 13 C value at “Crooked Creek.” This implies that T. tilletti had a Δ 13 C d-e of ~13-14‰ similar to modern mammalian C 3 consumers (Koch, 1998) and that Tenontosaurus tilletti had a preferred C 3 -based dietary resource that was available in both ancient Wyoming as well as ancient Texas. This has the added implication that dietary subsidization with marine resources is more likely the driver for higher δ 13 C enamel values in the other coastal taxa rather than an environmental cause.

Lepisosteidae indet

The specimen OMNH 64911, a Lepisosteidae indet. scale has an δ 13 C enamel value of –5.30‰ VPDB from Ostrom Unit VII or the upper Himes Member of the Cloverly Formation (Ostrom, 1970; D’Emic et al., 2019). This specimen may represent slight marine resource subsidization of the Cloverly Formation ecosystem. The δ 13 C value of this specimen is within the range or exceeds that of the Antlers, Woodbine, Wayan, and Dunvegan formation averages. OMNH 64911 appears to potentially be a marine-freshwater transient gar and indicates this Cloverly Formation taxon might have participated in marine ecosystems. Indeed, today, streams with anadromous fish, feature aquatic organisms with a slightly higher δ 13 C value because of anadromous fish migrations (Bilby et al., 1996; Ledogar et al., 2018). Like anadromous coho salmon in Washington state, USA (Oncorhynchus kisutch) and anadromous sturgeon in New York state (Acipenser sp.), higher δ 13 C values are transported from marine environments to inland environments through trophic food-chain interactions (δ 13 C muscle = –17.9 ± 0.3‰ VPDB and δ 13 C collagen = –15.3 ± 0.4‰ VPDB, respectively.) In contrast, isotopic compositions from freshwater cutthroat trout of the same river tributaries: δ 13 C muscle = –27.5 ± 1.4 VPDB (Bilby et al., 1996; Ledogar et al., 2018) are much lower in isotopic value.

Another explanation for an increased δ 13 C value is that aquatic plants, through modified C 3 photosynthetic pathways, assimilate carbon with higher δ 13 C values from aqueously dissolved bicarbonate (HCO 3 -) (Keeley and Sandquist, 1992; France and Schlaepfer, 2000; Tütken, 2011). Additionally, microbial respiration from decomposition in humic wetlands can also result in a change in the bulk δ 13 C value within the streams/rivers (France and Schlaepfer, 2000). While this may explain the increased δ 13 C value, this does not explain the very low δ 18 O enamel value of OMNH 64911. The δ 18 O value of vertebrate bioapatite is primarily controlled by: 1) temperature of formation and 2) isotopic composition of precipitation, which in this continental environment is dictated by latitudinal location or high altitude influence (Gat, 1996; Bowen and Wilkinson, 2002; Suarez et al., 2021). The δ 18 O enamel value from OMNH 64911 is significantly lower relative to the δ 18 O enamel of other taxa from the Cloverly Formation and could suggest its geographic origin to either have been from a high-latitude boreal seaway with higher δ 18 O water value or from a water body influenced by snowmelt/heavily fractionated precipitation from high altitudes but with very high ambient δ 13 C value as well that did not affect any of the other Cloverly Formation taxa. Additional elemental analyses were performed to help determine which explanation is more likely.

As mentioned above, Sr/Ca ratios observed in the composition of fish scales and otoliths can be used to determine the geographic source of a fish (Courtemanche et al., 2005; Seeley et al., 2015; Daugherty et al., 2019). The average Sr/Ca ratio of OMNH 64911 in mmol/mol is ~2.85 ± 0.19 (minimum=2.56, maximum=3.45). This is comparable to marine derived fish today (Seeley et al., 2015, 2017). The same analyses were conducted on two Lepidotes (?) teeth from the Antlers Formation, PMNS 2007-11-070 (formerly DMNH 2007-11-070), which had average Sr/Ca ratios of ~5.98 and ~5.02 mmol/mol, which are well within marine values. Although the Sr/Ca ratio of OMNH 64911 is not as high as that of the Antlers Formation fish, the ratio is still within the range of marine derived fish today and indicates high salinity concentration (Macdonald and Crook, 2010; Seeley et al., 2015; Daugherty et al., 2019).

OMNH 64911 may have been transient between freshwater and seawater and/or consumed anadromous fish with high δ 13 C value and Sr/Ca ratios and then subsequently acquired marine-sourced isotopic and elemental signatures (Jaecks et al., 2016; Daugherty et al., 2017, 2019). Regardless, both hypotheses confirm a marine influence for the Lepisosteidae and Lepidotes (?) fish of the Cloverly and Antlers formations. If OMNH 64911 is transient and deposited within its life habit range, it was likely within 60 kilometers of its home river/water body (Daugherty et al., 2019; Gray and Brandon, 2025). Its high δ 18 O, that is not shared with any other Cloverly Formation taxa, including other fish, supports an interpretation of transient behavior over an acquired isotopic and elemental composition and implies the upper Himes Member of the Cloverly Formation was deposited within 60 river kilometers of an estuary. Thus, anomalous δ 13 C values of these Cretaceous deposits are the result of marine-based resources subsidizing the base of the trophic system and suggest the start of establishment of a terrestrial-marine connection in the Upper Cloverly Formation prior to seaway incursion and deposition of the Sykes Mountain Formation.

The Cloverly Formation does not indicate significant marine influence based on lithologic characteristics and, hence, the bulk organic matter within this ecosystem is likely to be near the average δ 13 C value of the local plant life. The Antlers, Woodbine, Dunvegan, and Wayan Formations do have indications of coastal facies such as marine fossils, interbedded marine layers, and strandplain deposits and likely have a decoupling of local bulk organic δ 13 C value of the soil as well as the δ 13 C value of bioapatite of vertebrates in those ecosystems. Values of enamel carbon isotopic composition that suggest dietary supplementation with modified C 3 plant makeup imply there is a decoupling in the Δ 13 C of bioapatite to terrestrial bulk organic matter. Within the Cretaceous, this is likely along coastlines and major rivers with marine connections. MS of coastal deposits in the Cretaceous has not been considered in Cretaceous ecosystems despite numerous studies with ‘anomalous’ carbon sources (Stanton Thomas and Carlson, 2004; Fricke et al., 2008; Fastovsky et al., 2020; Cullen et al., 2023). We have demonstrated the likely presence of MS via organic material with higher δ 13 C value into terrestrial ecosystems of several ‘coastal’ deposits along the margins of the Western Interior Seaway as well as the Gulf of Mexico. The Antlers, Dunvegan, Wayan, Woodbine formations, and the Mussentuchit Member of the CMF are coastal deposits with fossil assemblages that were subsidized with higher δ 13 C value marine plant/macroalgae matter resulting in higher relative average δ 13 C bioapatite value.

The average δ 13 C value of CO 2 in the ancient atmosphere was not high enough to explain this offset. Neither the average δ 13 C value of the global atmosphere, a Δ 13 C d-e of 18‰ VPDB of hadrosaurs or 16‰ for sauropods, or small-scale microenvironmental factors are sufficient enough to explain the high δ 13 C value of the average δ 13 C value of bioapatite of so many taxa. MS of coastal environments provides a mechanism for whole ecosystem increases in bioapatite δ 13 C values.

While not significantly marine influenced, the Cloverly Formation includes hints of subsidization with the presence of a gar scale (OMNH 64911) with very high δ 13 C value and low δ 18 O value implying a marine boreal seaway source or dietary subsidization of prey from a boreal seaway. The gar scale also has an average Sr/Ca ratio which indicates a marine/estuary life habit tying the high δ 13 C value to a marine/estuary source habitat. Gar fish teeth from the Antlers Formation also have Sr/Ca ratios in line with modern marine fish composition but have much higher δ 18 O values indicating they are likely derived from the Gulf of Mexico.

Tenontosaurus tilletti appears to prefer non-subsidized C 3 plants in both the Cloverly and Antlers formations but the paucity of specimens requires further investigation. Organisms with dietary preferences of non-marine sourced food are frequently found in modern coastal ecosystems. The Δ 13 C d-e of the Cloverly Formation taxa from bulk organic matter indicates a much smaller Δ 13 C d-e for average herbivorous dinosaurs of +13-14‰ in line with modern mammalian/bird C 3 Δ 13 C d-e rather than the previously published +16 and +18‰. These Δ 13 C d-e values are the result of incorporation of marine organic matter rather than a unique physiological effect. Geologic, lithologic, and geographic information or other isotope systems are necessary to differentiate coastal from strictly terrestrial carbon.

Altogether, marine macroalgae and plants may represent alternative food resources for coastal organisms and are/were important resources that subsidized a significant portion of the terrestrial fauna particularly among the herbivorous dinosaurs. Some consumption of marine resources may be ubiquitous for some niches such as for neosuchians, coastal fish, turtles, and carnivores where opportunistic behaviors and aquatic living are expected however, cross-boundary resources in ancient ecosystems must not be ignored in paleoecological studies and is an under-explored area for future research.