An update on Aenocyon dirus in the interior of North America: new records, radiocarbon dates, ZooMS spectra, and isotopic data for an iconic late Pleistocene carnivore

Mauer, Iowa

Joe and Buertess Beals collected this partial A. dirus cranium in August 1959 from the now-flooded Mauer sand-and-gravel pit on the Boyer River in Crawford County, Iowa (Figs. 1B, 1C). The find-spot is 0.3 km north of the small town of Dunlap (WGS84 41.86N, 95.60W). The specimen predates the Last Glacial Maximum (LGM) (26,500–19,000 cal B.P.) (Clark et al., 2009) by about 2,500 years, and is currently the oldest directly dated A. dirus east of the Rocky Mountains. It is curated at the Sanford Museum and Planetarium (SMP), Cherokee, Iowa (catalog number 149-59-Z). The SMP’s paleontological locality number is 13CF4.

Villisca, Iowa

The A. dirus radius was collected by Jim McClarnon in May 2019 in shallow water resting on the bed of the West Nodaway River, 6 km north of the small town of Villisca, Montgomery County, Iowa (WGS84 40.97N, 94.98W) (Figs. 1B, 1C). Direct dating suggests it likely eroded from an upstream exposure of Noah Creek Formation, an extensive, thick unit of late Pleistocene sands-and-gravels (~17,000–13,000 cal B.P.) (Bettis & Kemmis, 1996, 23–24; Quade, 2006; Tassier-Surine et al., 2012). The specimen is curated in the Paleontological Repository, Department of Earth and Environmental Sciences, University of Iowa (SUI), Iowa City, Iowa. The catalog number is SUI-149737.

Peccary Cave, Arkansas

Peccary Cave is situated in Ordovician limestone in the southern Ozarks, near the confluence of Ben’s Branch and Cave Creek, a tributary to the Buffalo River, in Newton County, northwestern Arkansas (WGS84 35.92N, 92.97W) (Figs. 2A, 2B). Excavations in 1967–1969 (University of Arkansas, UA) and 1971–1972 (University of Iowa) yielded a large, taxonomically diverse collection of extinct and extant vertebrate remains. The identified small mammals are curated at SUI, while UA retains the balance of the collection. Only the last four digits of UA accession numbers are used in the text and illustrations because the full numbers are cumbersome (e.g., UA-67-300-075-4370 = UA-4370).

The original cave entrance, through which an adit (horizontal tunnel) was dug to facilitate paleontological investigations, was located 35 m northeast of the main chamber, flush with the valley-floor of Ben’s Branch, and measured 9 m wide and 2 m high (Quinn, 1972, 91) (Fig. 2C). Today this entrance is 8 m above the valley-floor. The main chamber measures 25 m east/west and 10 m north/south. It extends in a southwesterly direction into a relatively long, narrow tunnel (Fig. 2D), and in a northwesterly direction into a smaller room measuring 10 m east/west and 5 m north/south. Several low passages branch from the smaller room and encircle limestone columns. Ceiling height ranges from about a meter to places where the floor and ceiling nearly converge. The bone-bearing stratum extends to a depth of 50–60 cm and consists of a highly churned mixture of bones and teeth, Platygonus compressus droppings, Mesodon shells, fine-grained alluvium (clay, silt, and sand), and angular rock (≤50 mm in diameter) (Davis, 1967, 1969; Quinn, 1972; Semken, 1984, 407). The churning is likely the result of a slow rate of sedimentation and bioturbation by P. compressus and Erethizon dorsatum. Given that Mesodon does not inhabit caves (Hubricht, 1985, 42–45), it undoubtedly was transported as flotsam in floodwaters that breached the cave entrance and flowed into the main chamber.

Previously, Davis’ (1973) work on the reptiles and amphibians, along with the studies by Semken and colleagues on the small mammals (Graham & Semken, 1976; Hallberg, Semken & Davis, 1974; Megivern, 1982; Semken, 1984; Semken, Graham & Stafford, 2010; Stafford & Semken, 1990; Stafford et al., 1999), accounted for the substantive research on the collection. More recently, Wilson and Hill (Wilson, 2017; Wilson & Hill, 2020) analyzed the P. compressus assemblage. It includes 100 adult and subadult animals, as well as 14 fetuses in various stages of development. Of the other extinct medium- and large-sized taxa, the A. dirus sample is the largest, totaling 26 identified specimens representing at least 3 individuals. This sample is described and discussed here for the first time. Three dates on A. dirus and five dates on P. compressus suggest most, if not all, remains of these taxa accumulated during the LGM.

Systematic Paleontology

Class Mammalia (Linnaeus, 1758)

Order Carnivora (Bowdich, 1821)

Family Canidae (Fischer von Waldheim, 1817)

Genus Aenocyon (Merriam, 1918)

Aenocyon dirus (Leidy, 1858)

dire wolf

Referred material

Villisca: radius, SUI-149737.

Mauer: cranium, SMP 140-59-Z.

Peccary Cave: humerus, UA 67-300-015-1132; cuneiform, UA 67-300-240-1911; scapula, UA 65-158-036-2805; metacarpal V, UA 67-300-254-2958; metacarpal V, UA 67-300-051-2959; metacarpal V, UA 67-300-175-2960; metacarpal/metatarsal V, 6 UA 7-300-127-2961; metacarpal III, UA 67-300-194-2962; calcaneus, UA 67-300-239-2964; metacarpal III, UA 67-300-198-2965; second phalange, UA 65-158-014-3701; C1, UA 67-300-115-3737; third phalange, UA 67-300-182-3795; P4, UA 67-300-300-4362; m1, UA 67-300-30?-4363; m1, UA 67-300-018-4364; m1, UA 65-158-019-4365; P4, UA 67-300-210-4368; c1, UA 67-300-075-4369; C1, UA 67-300-034-4370; C1, UA 67-300-184-4371; C1, UA 67-300-360-4372; M1, UA 65-158-036-4374; C1/c1, UA 67-300-002-4375; m1, UA 67-300-281-4378; C1, UA 65-158-018-4742.

Descriptions

Villisca. An anatomically complete left radius (Fig. 3). Union of the proximal and distal articular epiphyses is complete. Physical condition and morphological integrity are good. Fluvial transport via traction and saltation has exposed cancellous bone on the margins of the proximal end as well as slightly rounded or polished topographic highs. A small patch (2 cm × 2 cm) of cortical cracking and flaking on the midshaft on the cranial surface suggests a period of subaerial exposure earlier in its taphonomic history.

Mauer. The preserved portion, a neurocranium, is in fair physical condition (Fig. 4). The foramen magnum and left auditory bulla are partially filled with coarse sand, and the foramen magnum also with a subrounded pebble. These particles are likely traces of the glaciogenic alluvium that stored the specimen in secondary context until discovery. Uniform staining of the fracture edges and intact portions indicate separation of the preserved portion and other cranial parts occurred prior to exposure and recovery. Intact elements include the frontals, parietals, temporals, occipital, basisphenoid, and pterygoids. The external occipital protuberance of the sagittal crest is missing. The zygomatic arches are represented by broken nubs. Both external acoustic meatus are present; the fragile outer (inferior) surfaces of the tympanic bulla are missing. Following a system to describe suture closure in Homo sapiens crania (Milner & Boldsen, 2013, 12), the fronto-parietal and parieto-squamosal sutures are partially obliterated, while the parieto-supraoccipital, interfrontal, and basioccipital-basisphenoid sutures are obliterated.

Peccary Cave. The assemblage includes 14 complete teeth or tooth fragments and 12 postcranial specimens. These specimens are registered in Table 1 and shown in Fig. 5. Overall, rodent gnawing and calcium carbonate (CaCO3) accretion have significantly impacted the morphological and metric integrity of the specimens, and have also obscured cortical surfaces. This characterization applies to the entire faunal collection. Many long bones and metapodials have been sculpted to cylinders, carpals and tarsals to amorphous nodules, and teeth to crowns (sensu Ball & Davis, 1993, 124). E. dorsatum, a notorious gnawer, is inferred to be the main culprit (Curtis & Kozicky, 1944, 138; Labanoski, 2017; Pokines et al., 2017, 51; Woods, 1973, 4). Mineral accretion ranges from small, isolated patches and nodules to clumped masses that encase entire specimens.

Morphological diagnoses

Villisca radius. The length of the radius falls within the range of A. dirus from several caves in Missouri (Table 2), all of which surpass the longest example in a sample of 740 specimens from RLB by 1–2 cm (Fig. 6A). In this attribute, it exceeds C. lupus. Bivariate arrangements of articular end measurements show similar separation of the taxa (Figs. 6B, 6C). The Missouri specimens lack analogous measurements to be displayed in these plots; however, the available measurements are similar to those in the Villisca specimen (Table 2). Also, the middle groove on the cranial aspect of the distal end, which houses the tendon for the m. extensor carpi radialis (Evans, 1993, 191), is shallow and broad, similar to that described in A. dirus at RLB and dissimilar to that in C. lupus (Merriam, 1912, 237). These comparisons coupled with its overall robustness and relatively faint middle groove are the basis for assigning the Villisca radius to A. dirus.

Mauer cranium. Direct comparison with the female Alaskan C. lupus cranium and indirect, online comparison with a three-dimensional scan of a C. rufus cranium indicates the temporal ridges, sagittal crest, and zygomatic remnants in the Mauer specimen are far more robust than in the comparative taxa. The external occipital protuberance, of which only the base is preserved, was apparently massive and probably extended well beyond the occipital cliff. The basioccipital is also exceptionally broad with massive tympanic bullae. The occipital region closely matches that described in A. dirus at RLB (Merriam, 1912, 226). In posterior view, the lateral occipital ridges converge dorsally to form a relatively sharp, compressed lambdoidal crest. In C. lupus, the lateral ridges gradually converge to form a relatively round, broad lambdoidal crest (Fig. 4). Finally, standard width measurements across the postorbital processes and at the postorbital constriction (Table 3) fall within the ranges in A. dirus (Figs. 7A, 7B). In these attributes, there is minimal overlap with C. lupus and no overlap with C. rufus. These findings are the basis for assigning the Mauer cranium to A. dirus.

Peccary Cave. Morphological diagnoses rely on metric comparisons of several specimens. There are two P4s: a complete tooth (UA-4362) and a mesiolingual enamel fragment (UA-4368). The complete tooth is small compared to A. dirus, and in fact, it approximates the size of the tooth in C. lupus (Fig. 8A). There is no overlap in mesiodistal length with C. rufus, except for one female from southeast Missouri (UNSM 244528). However, the tooth displays a reduced deuterocone, as does the enamel fragment, which is consistent with A. dirus (Merriam, 1912, 228). Only the talonid portion of the M1 (UA-4374) is present, and notably as in A. dirus, the hypocone ridge does not reach the anterior side of the protocone (Merriam, 1912, 231). Of the four m1s, three are fragments (UA-4363, -4364, -4365) and one is a complete tooth (UA-4378) covered in a thin coating of CaCO3. Except for the overall robusticity of the fragments and the mesiodistal length of the complete specimen, they do not retain or expose features to distinguish them from C. lupus. Although the length of the complete specimen (35.7 mm) (UA-4378) is perhaps a millimeter or two greater than its actual length because of CaCO₃ encrustation, it is larger than those in C. lupus and C. rufus and falls within the A. dirus size range (Fig. 8B). The buccolingual breadth of the tooth is too encrusted in CaCO3 for reliable measurement.

Several postcranial specimens offer useful diagnostic information. The only standard measurement obtainable on the scapula (UA-2805) is the depth of the glenoid fossa. It minimally measures 37.7 mm due to minor damage at both measurement points. Still, the feature is larger than in C. lupus and in line with A. dirus (Fig. 8C). The same size relationships occur in the midshaft breadth of the humerus (19.6 mm) (UA-1132) (Fig. 8D). Two metacarpi V (UA-2958, UA-2960) are similar in length to those from several Missouri caves and the longest specimens from RLB (Table 4). They are about a centimeter shorter than one from Cherokee Cave (Fig. 8E). Moreover, the calcaneus (UA-2964) is about the same length as those from Hawver Cave, Ingleside, Powder Mill Creek Cave, and the largest ones from RLB, all of which are substantially shorter than the Carroll Cave specimen (Table 5) (Fig. 8F). The element is typically smaller in C. lupus.

The weight of metric and nonmetric observations on the Peccary Cave large canid remains is consistent with A. dirus. This diagnosis is extended to the remainder of the sample.

ZooMS diagnoses

We attempted to extract collagen from both Iowa specimens, along with the three directly dated specimens (UA-2964, -2958, -2959), the humerus (UA-1132), and the scapula (UA-2805) from Peccary Cave for ZooMS. Only the Iowa specimens and the Peccary Cave scapula produced usable substance. Figure 9 profiles partial MALDI-TOF MS spectra of the tryptic digests of bone collagen from these specimens and modern C. lupus and C. latrans (Data S6). The paleontological samples are identified as canid based on the presence of the ɑ2 484 peptide at 1453.7 m/z, the ɑ2 793 peptide at 2,131.1 m/z, and the ɑ1 586 peptide at 2,853.4 m/z. The modern canids are well characterized (Buckley et al., 2009) and have ɑ2 978 peptide peaks at 1,210.7/1,226.7 m/z. The Villisca, Mauer, and Peccary Cave spectra are indistinguishable from each other and importantly, do not exhibit the ɑ2 978 peak of C. lupus and C. latrans. In A. dirus, that peptide appears to have shifted to 1,180.6/1,196.6. A difference in the ɑ2 978 peptide is also suggested by the amino acid sequence published by Perri et al. (2021) for A. dirus.

ZooMS thus indicates the Mauer cranium, Villisca radius, and Peccary Cave scapula are not Canis. This finding excludes the possibility of C. lupus, which occurs at Natural Trap Cave, Wyoming (Meachen, Brannick & Fry, 2016), and C. rufus, which occurs in the regional paleozoological record (Nowak, 2003, Fig. 9.7). The only viable alternative taxonomic candidate for them is A. dirus, which is also assigned to the remainder of the Peccary Cave large canid remains.

Radiometric and isotopic results

AMS radiocarbon dates, isotopic results, and related data for both Iowa specimens and three Peccary Cave specimens are provided in Table 6. The C:N ratios (3.1–3.1) suggest the measured samples are not severely degraded or affected by exogenous contaminants, indicating that the dates and isotopic results are likely reliable (Ambrose, 1990, 447; Talamo et al., 2021, 63-64; van Klinken, 1999, 691). All of the dates fall within MIS 2 (29,000–11,700 cal B.P.). Mauer dates to 29,040–28,410 cal B.P., Peccary Cave to 25,350–21,405 cal B.P., and Villisca to 14,325–14,075 cal B.P.

Occurrences

Taking into consideration minor changes to Dundas’ (1999) inventory, 26 records published after his census, the addition of 13 records he overlooked or omitted from his study, and the two new Iowa occurrences, there are now 166 A. dirus records in southern North America (Fig. 1) (Data S7). Roughly half (n = 70) of these records—including Mauer and Villisca—are not reported in the Neotoma Paleoecology Database.

Notable additions and updates to the A. dirus record include:

1. A nearly complete cranium and several other elements, including measurements, from Burnham, Oklahoma (Czaplewski, 2003).

2. An update on the partial skeleton from Marlow, Oklahoma (Cifelli, Smith & Grady, 2002).

3. Notice of a complete skeleton from Megenity Peccary Cave, Indiana, belonging to an adult male that had fallen into a 6-m steep-walled pit and became trapped (Richards, 1995, 90; Richards & Whitaker, 1997, 149).

4. Measurements on the complete skull (cranium + mandibles) from Ardis, South Carolina (Sanders, 2002, Table 3).

5. Descriptions of material from Medicine Hat (Surprise Bluff), Alberta (Reynolds et al., 2023), Kincaid Shelter, Texas (Johnson & Moretti, 2022), Natural Trap Cave, Wyoming (Meachen, Brannick & Fry, 2016; Redman et al., 2023, 44), Murray Springs and Lehner Ranch, Arizona (Hemmings, 2007, 103; Saunders, Baryshnikov & Seymour, forthcoming), and Peccary Cave, reported here.

6. The first records for Iowa, reported here, and for Nevada, from Tule Springs (Scott & Springer, 2016).

7. The addition of the taxon to faunal lists for Zesch Cave, Texas (Sagebiel, 2010), and Ladds, Georgia (Nowak, 2002, 107).

8. Another specimen, an m1, from Stratum 2/V at Vero, Florida (Adovasio, Hemmings & Vento, 2024, 83; Hemmings et al., 2018, 87).

9. Direct AMS dates from Natural Trap Cave, Wyoming, and Guy Wilson Cave, Tennessee (Perri et al., 2021).

10. Development of a series of ≥76 direct AMS dates for RLB (Friscia et al., 2008; Fuller et al., 2014; Fuller et al., 2015; Fuller et al., 2020; O’Keefe et al., 2023).

11. Age-bracketing of the material from Hall’s Cave, Texas (Seersholm et al., 2020; Waters et al., 2021, 9).

12. Reassignment of C. mississippiensis to C. lupus, with the provenance being the lead region, and verification of the Blue Mounds, Wisconsin C. lupus record (Addendum).

13. Placement of the provenance of a mandible reported by Merriam (1912, 221–222, 242–243) and Hibbard (1939, Table 1) at Sheridan, a ghost town in Logan County, Kansas (Addendum).

These additions and updates do not substantially change the general character of the record reported by Dundas (1999). The species ranges in time from as early as MIS 19 at Fairmead Landfill and Irvingtonian 2, California (Tedford, Wang & Taylor, 2009, 132; Trayler et al., 2015, 132) to MIS 2. Two-thirds (n = 112) of the occurrences date to MIS 2, MIS 2–3, and MIS 3 (Fig. 10). Geographically, it spans from the Pacific Coast to the Atlantic Coast, and from El Cedral, Mexico, in the south to Medicine Hat, Alberta, in the north (Fig. 1). Actual absence, non-preservation, and inadequate sampling account for the spotty distribution of occurrences in some regions (Grayson, 1981). Notable clusters of occurrences are found in Florida (n = 32), California (n = 29), and in and around the Ozarks (n = 13). The southern Great Plains also contains a relatively large number of occurrences. The Ozark record represents the largest concentration of A. dirus material in the interior of southern North America, encompassing a variety of find-types, from a few bones or teeth to nearly complete skeletons. State-line records from locations such as Natural Trap Cave in northern Wyoming, and Cherokee Cave and Herculaneum Fissure in Missouri, strongly suggest the taxon ranged into Montana and Illinois, for example (Merriam, 1912, 222).

Biogeography

The distribution of occurrences highlights generalist adaptations in A. dirus. The taxon apparently thrived in various environments with different prey options. For instance, in California over the past 750,000 years (MIS 2–19), the combination of stable isotope data and faunal associations suggest the taxon preyed mainly on Camelops, Bison, Equus, and Hemiauchenia in warm, xeric settings such as open prairies, mixed woodlands, and scrublands (DeSantis et al., 2019; Fuller et al., 2014; Fuller et al., 2020; Trayler et al., 2015; O’Keefe et al., 2023). Similarly, at Cutler Hammock, Florida (late MIS 2), it lived a warm environment transitioning from relatively closed mesic conditions to open xeric conditions. Consequently, its diet shifted from browsers such as M. fossilis, P. compressus, and Odocoileus to grazers such as Bison, Equus, Hemiauchenia (Munson, 2022, 31–32; cf. D’Amo, 2001, 39–43). The Mauer, Villisca, and Peccary Cave records provide information from the interior of North America, within a continental environmental context.

Proxy evidence from Logan Quarry, on the Boyer River (Fig. 1C), about 30 km downstream from the Mauer find-spot, indicates the animal inhabited an environment that was transitioning from full boreal to full glacial conditions (Baker et al., 2009, Table 1). Fifteen thousand years later, during the early Bølling-Allerød Chronozone (14,640–12,850 cal B.P.) (Byun et al., 2021; Gonzales & Grimm, 2009, 242), the Villisca animal inhabited a boreal parkland (Rhodes & Semken, 1986, 109).

The environment A. dirus inhabited is clearer at Peccary Cave. During the LGM, the Ozarks were covered in a Picea- and Pinus-dominated boreal forest interspersed with some deciduous hardwoods that has no modern analog (Jackson et al., 2000; King, 1973). The understory and forest openings were vegetated with Cyperaceae, Poaceae, and Artemisia shrubs and forbs, among other taxa. Palynological evidence from Cupola Pond, 200 km northeast of Peccary Cave (Fig. 2A), indicates this forest persisted until 15,000 cal B.P. (Jones, Williams & Jackson, 2017, Figure 7). Small mammal taxa from Peccary Cave also record a non-analog environment (Semken, Graham & Stafford, 2010; Stafford et al., 1999). As illustrated in Fig. 11, the majority of small mammals date to approximately 20,000 cal B.P. (Data S8), and significantly, the associated range maps demonstrate contemporaneity of several taxa with modern distributions that do not overlap, specifically, northerly tundra/boreal taxa and several southerly prairie/deciduous forest taxa. Compared to the modern situation, mean January and July temperatures were 15–20°C and 10°C colder, respectively, and there was ~40 mm less precipitation (Jackson et al., 2000, 500).

Diet

Leaving aside subadult and adult proboscideans, accounts of prey selection and hunting in C. lupus (Mech, Smith & MacNulty, 2015; Peterson & Ciucci, 2003) reveal that any herbivore co-existing with A. dirus was not immune to predation. In western Iowa, an absence of direct dates on potential prey species makes it difficult to put A. dirus diet on solid ground at the present time. However, the list of practical candidates is short. It includes Bootherium, Megalonyx, and Cervalces, whose remains are relatively abundant in the regional fossil record (Barbour, 1931; Churcher, 1991; Churcher & Pinsof, 1988; Dulian, 1975; Hay, 1914, 78–79, 306; McDonald, 2012). P. compressus is known only from several locations in southwestern Wisconsin, northeast Iowa, and northwestern Illinois (Wilson & Hill, 2020, Figure 1), while its solitary counterpart, M. fossilis (Kurtén & Anderson, 1980, 296), has yet to be reported in the state. Months-old proboscideans were probably uncommon prey due to the unpredictable nature of the necessary circumstances for attacks to occur (see Van Valkenburgh et al., 2016).

Comparison of stable nitrogen isotope results obtained for A. dirus from Villisca and for S. fatalis from a sand-and-gravel operation 2 km north of Shenandoah (WGS 84 40.76N, 95.37W) (Fig. 1C) (Hill & Easterla, 2023, Table 1) casts light on the diet of these carnivores during the Bølling-Allerød in southwest Iowa. Briefly, studies on the diet of predators that consume mostly vertebrate prey indicate the difference between the isotopic composition of predator and prey bone collagen directly relates to the isotopic enrichment that attends each trophic step. The δ¹⁵N value of the collagen, in particular, increases 3–5‰ with each trophic step, from primary consumers to top consumers (Bocherens & Drucker, 2003; Schoeninger & DeNiro, 1984). The S. fatalis result (δ¹⁵N = 8.2‰) (Hill & Easterla, 2023, Table 1) is a trophic step higher than the A. dirus result (δ¹⁵N = 5.8‰) (Table 1). While the absence of complementary isotopic data on potential prey constrains what can be inferred from these values, it tentatively suggests the two taxa were possibly not in regular competition for the same prey and occupied different niche spaces, as appears to be the case at RLB (DeSantis et al., 2019, 2489). This general idea is supported by compelling evidence that S. fatalis was a solitary ambush hunter, whereas A. dirus hunted in packs (Brown et al., 2017).

In the Ozarks, P. compressus was ideal prey, an idea alluded to decades ago by Hood & Hawksley (1975, 14). With adults weighing 50–75 kg (Guilday, Hamilton & McCrady, 1971, 291), it was probably the most common medium-sized herbivore in many late Quaternary faunal communities (Wilson & Hill, 2020). To this point, actualistic work in savanna and temperate ecosystems reveals a high level of fidelity between the number of a taxon’s bones on the landscape and its living abundance (Miller, 2011; Miller, 2012; Western & Behrensmeyer, 2009). If this relationship holds for closed settings such as Peccary Cave, then P. compressus (MNI = 100) outnumbered A. dirus (MNI = 3) outside the cave by a wide margin, as would be expected given fundamental trophic linkages (Pimm, Lawton & Cohen, 1991). Second, while its defensive weaponry—large canines or ‘tusks’—conferred some protection from predation, P. compressus was far less capable of inflicting serious counter-injuries during attacks than other coexisting megaherbivores. Compared to P. compressus, these animals weighed hundreds of kilograms and possessed potentially lethal horns, antlers, clawed forelimbs, and powerful, rapid-fire hooves to combat take-down attempts (Brown et al., 2017; Van Valkenburgh & White, 2021).

Following the same analytical tack as above, stable isotope analyses on A. dirus and P. compressus from Peccary Cave support the existence of this inferred predator-prey relationship between 25,000 cal B.P. and 21,000 cal B.P. Isotopic data for A. dirus and P. compressus are provided in Tables 6 and 7. The expected relationship between results for the taxa holds true (Fig. 12). The δ¹⁵N_coll values for A. dirus (range 6.1–8.0‰) are 3–5‰ higher than those for P. compressus (range 3.7–4.8‰). While another prey taxon such as Bootherium or Megalonyx might be driving the A. dirus values, P. compressus is the most probable candidate behind them in this regional LGM faunal community.

Aenocyon dirus and Platygonus compressus use of Peccary Cave

The radiocarbon dates indicate the use of the Peccary Cave by P. compressus and A. dirus between 25,000 and 21,000 cal B.P. (Fig. 11), which apparently bookends the period when these animals were able to navigate the passageway leading to the main chamber (Fig. 2C). Binford (1981, 200) notes a similar cave-use situation in Alaska involving C. lupus and Ovis dalli. By 21,000 cal B.P., the passageway was partially choked with poorly sorted alluvium from overbank flooding of Ben’s Branch, so much so that it appears to have effectively blocked these large animals from entering the cave. This opening, or an unknown one, remained accessible to small mammals until 19,000 cal B.P. when all access points were apparently sealed. Small mammals somehow gained entry again between 13,000 and 10,000 cal B.P. until that opening closed. Later, in the short passage off the south wall of the main chamber, a 10 m chimney opened and a 4.26 m tall debris cone beneath it began to build (Fig. 2C) (Ball & Davis, 1993; Davis, 1969, 196). Several middle Holocene radiocarbon dates on Neotoma floridana (Semken, Graham & Stafford, 2010, Table 1) may approximate the timing of chimney formation.

The high MNI, presence of fetuses and droppings, and observations of cave use in Pecari tajacu (Sowls, 1997, 67, 130–133) indicate regular use of the cave by P. compressus, with one individual dying inside the cave approximately every 45 years (Wilson & Hill, 2020, 8). The rate and duration of visitation probably increased during the winter as animals sought protection from the bitter cold. For A. dirus, at least three adult animals died inside the cave over a period of four thousand years. Two of these individuals were relatively old. In Stiner’s (1994, Fig. 12.3) nine-cohort eruption-and-wear sequence for canids, the two right, partial m1s are heavily worn. As best as can be determined, one specimen appears to belong to group VII (UA-4364), and the other to group VIII (UA-4363). The remaining m1, identified as a left buccal fragment (UA-4365), cannot be assigned to a group due to the lack of a readable occlusal surface. However, this specimen was fully erupted and appears to have belonged to a younger adult, relative to those represented by the two right m1s. These observations lead us to infer that the cave periodically functioned as a refuge for aged and infirm animals. Most likely, this mode of accumulation also explains the presence of partial skeletons with heavily worn teeth found at places such as Zoo Cave, Missouri (Hood & Hawksley, 1975, 28), Powder Mill Creek Cave, Missouri (Galbreath, 1964, 235), and Marlow, Oklahoma (Cifelli, Smith & Grady, 2002, 94).

Granting some parallels with C. lupus, the evidence does not point to its use as an A. dirus den. While den type, location, and sociality are thoroughly documented in C. lupus (Packard, 2003, 35–51), actualistic research on the bone debris that accumulates in and around C. lupus dens to aid in interpreting paleozoological patterns is sparse but revealing. Foremost, prey remains are rare inside dens but can be relatively common on entrance aprons, and second, these assemblages are “dominated by (heavily gnawed) skulls, teeth, feet, and lower limbs for the large and moderate-size mammals (i.e., multi-taxic)” (Binford, 1981, 202; Haber, 1977, Appendix 11; Haynes, 1980, 1982; Mech & Packard, 1990; Murie, 1944, 31; Packard, 2003, Figure 2.4; Peterson, 1977, 107; Young, 1964, 101–102). As well, reuse of bedrock overhangs, nooks, and crannies for denning is common in places where these features are not common and, consequently, prey-bone accumulations at these places are relatively large and time-averaged (Haber, 1977, 302). Finally, since whelping typically occurs in dens (Fuller, Mech & Cochrane, 2003, 183–184; Peterson, 1977, 112; see also Stiner, 1994, 286–287, 316–319), sometimes they contain puppy remains (Milideo, 2015, 39).

None of these expectations are met at Peccary Cave. Neonatal or very young A. dirus remains are not present. Differential preservation cannot account for this absence since fetal P. compressus bones and teeth are relatively common. Second, unambiguous signs of carnivore scavenging of P. compressus carcasses in the form of tooth scores, punctures, gastric etching, and destruction of articular ends of bones are limited to four specimens. Also, while the assemblage contains remains from other potentially viable prey species (i.e., multi-taxic), specifically, Cervalces (n = 3), Bootherium (n = 12), and Tapirus (n = 1), the representative specimens are isolated teeth. How these remains entered the main chamber is not clear, but it is notable that the Tapirus specimen is water-worn, as is a hand-sized chunk of proboscidean long bone. Several other proboscidean remains reported by Quinn (1972, 92) have not been relocated. (The Equus tooth mentioned by Quinn (1972, 92) is, as he suggests, modern in age.) Finally, the Ozarks contain 8,500 reported caves (Tennyson et al., 2017, 663) and probably many times that number of small crevices, fissures, and low overhangs. Given this, and at least over the short-term, A. dirus inhabiting the Ozarks probably did not often reuse natal dens, which when considered with the fact they are typically used only in the spring for several weeks in Minnesota (Fuller, 1989, Table 1; Mech, 2000, 78) and eight weeks in Alaska (Mech et al., 1998, 103), will make it exceedingly difficult to identify these places from faunal evidence (cf. Emslie & Morgan, 1995; Nye, 2007).

Sex and body proportions

The partial A. dirus skeleton from Powder Mill Creek Cave, Missouri, is inferred to belong to an adult female (Galbreath, 1964, 235), and similarly, the one from Zoo Cave is also mature and “quite small by Missouri standards” (Hood & Hawksley, 1975, 28). The partial skeleton from Marlow, Oklahoma, is also from a small, skeletally and dentally mature individual (Cifelli, Smith & Grady, 2002, 94). If the radii in these individuals approximate the size of adult females, then it seems probable the specimens from Brynjulfson Cave no. 1, Missouri, and the larger of the two from Bat Cave, Missouri, represent adult males (Fig. 6A). These specimens are also longer than in the “large male” (Galbreath, 1964, 241) from Carroll Cave, Missouri. All other things equal, it appears the Villisca radius belongs to a male. The greatest length is closer to that recorded for inferred males than inferred females. It is slightly longer than the longest male radius in the C. lupus sample. In prime physical condition, the Villisca animal may have approached 70 kg (Anyonge & Roman, 2006, 211), or about the same as exceptionally large C. lupus males (~80 kg) (Mech, 1970, 11–12). (In comparison, the measured live weight of the four heaviest males in the Great Lakes C. lupus sample is 44.4 kg.) This estimate aligns with observations that, overall, the morphotype found east of the Rocky Mountains was taller and more robust than its counterpart found west of the Rocky Mountains, and in turn, supports the inference that it was possibly faster and more agile (Kurtén, 1984, 226).

Timing of extinction

Outside of RLB, seven direct AMS radiocarbon dates from five localities are available for A. dirus. Figure 13 profiles these results in relation to those from RLB (n = 76) (Data S9). The RLB series minimally spans the period 50,000–13,000 cal B.P. (MIS 2 and 3), whereas the other dates minimally span the period 29,000–13,000 cal B.P. (MIS 2). Small sample size, dating failures (Cifelli, Smith & Grady, 2002, 94; MacFadden et al., 2012, 708; Perri et al., 2021), and non-preservation of directly dateable material in places such as Florida (Hulbert, Morgan & Kerner, 2009, 546; Morgan, 2002, 21) account for this discrepancy. Biostratigraphic ages of ~42,000 cal B.P. (MIS 3) from Medicine Hat (Surprise Bluff), Alberta (Reynolds et al., 2023, 940) and ~40,000 cal B.P. (MIS 3) from Burnham, Oklahoma (Czaplewski, 2003, 162; Wyckoff & Carter, 2003), for example, are good backfill, however.

The earliest date east of the Rocky Mountains is on the Mauer cranium, followed by three LGM dates from Peccary Cave (Table 6) and one from Natural Trap Cave, Wyoming (24,235–23,795 cal B.P.) (Perri et al., 2021), and then the Villisca date. The average of two assays on a tooth root from Guy Wilson Cave (12,965–12,755 cal B.P.) (Perri et al., 2021) is the current last appearance date for the taxon. Sixteen results from RLB fall between those from Villisca and Guy Wilson Cave, Tennessee. The tightly bracketed A. dirus remains from Hall’s Cave, Texas (14,700–12,900 cal B.P.) (Waters et al., 2021, 9) fall between the Guy Wilson Cave and Villisca dates. These observations indicate that A. dirus survived in geographically distant locations across southern North America until terminal extinction sometime after 12,800 cal B.P.

Reassignment of Joel Allen’s Canis mississippiensis to Canis lupus

In 1876, Joel Allen provisionally named Canis mississippiensis as an extinct species of large canid, reportedly double the size of Canis lupus and comparable in size to Aenocyon dirus, based on several bones from the lead region of contiguous parts of Wisconsin, Iowa, and Illinois (Fig. A1). As early as 1884 and as recently as 2023, confusion has surrounded the taxonomic affiliation and provenance of this taxon. It has been cited as a separate species (Bader & Techter, 1959, 2; Baker, 1920, 353–354; Hay, 1914, 484–487; Hay, 1923, 337), synonymized with A. dirus (Caro et al., 2023, 6; Damián, Xiaoming & Ascanio, 2022, 103; Dundas, 1999, Table 1; Kurtén, 1984, 219–220; Kurtén & Anderson, 1980, 171; Merriam, 1912, 220–221; Merriam, 1918, 532; Nowak, 1979, 115; Stevenson, 1978, 179; Tedford, Wang & Taylor, 2009, 114), and synonymized with C. lupus (Cope & Wortman, 1884, 11; Hoffmeister, 1989, 37). As far as is known, the species name has not been specifically assigned to other specimens. In regards to provenance, it has been reported as Illinois (Bader & Techter, 1959, 2; Cope & Wortman, 1884, 11), Wisconsin (West & Dallman, 1980), and Blue Mounds, Dane County, Wisconsin (Dundas, 1999, Table 1; Hay, 1923, 342; Kurtén, 1984, 219; Nowak, 1979, 115). Blue Mounds has also been mistakenly reported being located as in Jo Daviess County, Illinois (Anderson, 1905, 12). Despite their commendable efforts, Hay’s (1914, 484–487) and Nowak’s (1979, 115) attempts to resolve the confusion were only partially successful and introduced several errors. Given the direct impact of these issues on the biogeography of A. dirus, they were disentangled and reconsidered. C. mississippiensis is reassigned to C. lupus, with the provenance being the lead region. It is also been reaffirmed that Wyman (1862)’s canids from Blue Mounds are indeed C. lupus. The rationale is discussed in the following sections.

Historical context

In a tome on Wisconsin geology (Hall & Whitney, 1862), Jeffries Wyman (1862) assigned some canid bones from Blue Mounds, Dane County, Wisconsin (WGS 84 43.02N, 89.84W) to C. lupus (Fig. A1) In the next chapter, Joseph Leidy (1862, 424) reported on some material from Galena, Jo Daviess County, northwest Illinois (WGS 84 42.42N, 90.43W), 80 km southwest of Blue Mounds and 10 km south of the Wisconsin-Illinois border. He did not mention any canids then nor did he eight years later in another presentation of the fauna, which by that time included two additional taxa, Megalonyx and Bison (Leidy, 1870, 13). Both locations are within the lead mining region of Wisconsin, Iowa, and Illinois (Knox, 2019a; Walthall, 1981, 27–28), which may explain why Leidy (1862, 424), who was eminently qualified for the task, was asked to inspect the material and report his findings in a volume on Wisconsin geology in the first place. Why Leidy did the original (and subsequent) work, and not Wyman who was also eminently qualified, is unknown. Whatever the reason, Leidy was assuredly interested because Galena and the lead region generally had a track record for yielding extinct Pleistocene fauna, most notably to that point in time, the holotype of Platygonus compressus (LeConte, 1848a, 1848b, 1852).

Taxonomic and provenance revisions

‘Lead region’ is the most appropriate provenance for Allen’s large canid (C. mississippiensis). From whom and when he acquired the specimens is unknown, but obviously it was before publication in 1876. It is also unknown if they are from one or more places. As far as is known, no other canid material relevant to this discussion has been reported from the crevices and fissures around Galena. The provenance of Wyman’s two large canids (based on left mandibles) stays the same (i.e., Blue Mounds, Dane County, Wisconsin).

Allen’s lead region large canid (C. mississippiensis) and Wyman’s Blue Mounds large canids are almost certainly C. lupus. The lengths of the humerus and tibia in C. mississippiensis are about the same as those in smaller C. lupus males and larger females (Fig. A2). Both bones are also generally shorter than those in eastern A. dirus, especially the humerus. The tibia is about the length as inferred females from Bat Cave and Marlow. Both bones are slightly longer than the majority of those from Rancho La Brea. As well, Wyman identified the Blue Mounds canids as C. lupus. Although not stated specifically, it appears he directly compared the specimens to C. lupus in his diagnosis. While direct dating will be required to determine the precise temporal position of these specimens, it is notable that all other paleozoological occurrences of C. lupus in the region are Holocene in age (Fig. A1), a possibility that Hay (1914, 488) mentioned years ago.

Although the Blue Mounds and lead region canids can no longer be considered as relatively northern occurrences of A. dirus, the chances for future regional discoveries of the taxon seem decent. The Driftless Area (Fig. A1) (Knox, 2019b) contains many closed contexts (Alexander, 1980; Day, 1986; Martin, 1965, 93–97) that have not been systematically explored for late Pleistocene vertebrates (but see Foley, 1984; Slaughter & Jones, 2000; Thieling, 1973; Wallace, 2000, 2008; Widga et al., 2012). Additionally, the list of potential prey reported from the region includes Megalonyx (Leidy, 1870), Platygonus (LeConte, 1852; Palmer, 1974), Castoroides (Dallman, 1969), Bootherium (McDonald & Ray, 1989, 64); Cervalces and Rangifer (Josephs, 2005; Long & Yahnke, 2011; West, 1978), and Mammuthus and Mammut (Boszhardt et al., 1993; Widga et al., 2017). In the end, while fossil evidence is currently absent, it seems probable that A. dirus periodically ranged at least as far north as the lead region.

Kansas Canis lupus

The Kansas C. lupus skeleton (MCZ no. 268) is an enigma. It was originally cataloged as C. lupus occidentalis (= Alaskan wolf, Northwestern wolf) then changed to C. lupus nubilus (= buffalo wolf, Great Plains wolf) at an unknown later date. It was collected at Coyote, a now-abandoned railroad colony in Trego County, west-central Kansas, apparently by Allen and two assistants in early January 1872 on their collecting trip to the Great Plains and Rocky Mountains in 1871–1872 (Allen, 1916, 26–27; https://iiif.lib.harvard.edu/manifests/view/ids:49672077).

The humerus and tibia in the Kansas animal are only 5–10 mm longer than those in an exceptionally large adult male C. latrans from Iowa County, southwestern Wisconsin (UWZM 27383; live weight 15.5 kg, humerus = 165 mm, tibia = 194 mm), and 15–20 mm shorter than the lightest individual in the Great Lakes C. lupus sample (UWZM 27751, adult female; live weight 27.8 kg, humerus = 190 mm, tibia = 217 mm). These comparisons call into question the identification of the Kansas animal. It could actually be a really small female C. lupus. Alternatively, it may be a misidentified large C. latrans (= prairie wolf), or perhaps even a Canis hybrid. Whatever the case, it would be interesting to know for certain.

Provenance of the Aenocyon dirus mandible from Logan County, Kansas

Merriam (1912, 221–222, 242–243) reported an A. dirus mandible with p2-m3 from the Sheridan formation or beds of Kansas, in the collections of the American Museum of Natural History (FM-10391). Years later, Hibbard (1939, Table 1) questioned this provenance by prefixing it with a question mark (?) (“? Sheridan Formation, Kansas”), possibly because there is no stratum in the state with this name (Tolsted, Swineford & Buchanan, 1986). The American Museum online catalog indicates the specimen is from Logan County, Colorado. Specific provenance is not reported. Online searches for places and geological formations by the same name in Logan County, Colorado, failed to produce any promising leads. However, there is a ghost town named Sheridan in Logan County, Kansas (Snell & Richmond, 1966, 346–347), which incidentally, is located southwest of Sheridan County (Fig. A3). Knowing all is not well and good, the American Museum catalog appears to be incorrect. The most congruent provenance for the mandible is Sheridan (ghost town), Logan County, Kansas. Because the age is unknown, it is placed in the MIS 2-7 bin.