The Controversial Origin of Ferruginous “Coprolites”

Abstract

Ferruginous bromalites (coprolites and cololites) occur in enormous quantities in the Upper Cretaceous Whitemud Formation of Saskatchewan, Canada, and in Miocene deposits in Madagascar and southwest Washington, USA. The origins of these specimens have been the subject of diverse and often conflicting interpretations. This paper includes some discussion of other localities, but the main focus is on specimens from Wilkes Formation at Salmon Creek, Lewis County, Washington State, USA. This locality is notable because the geologic setting and paleoenvironment are well-established, and the purported bromalites can be observed in situ, providing stratigraphic and taphonomic information that is not available for the Canada and Madagascar locations. Past research at Salmon Creek has a curious history. Supporters of the coprolite interpretation have relied on Salmon Creek specimens collected by others. In contrast, field-based investigators have concluded that the extruded objects are probably pseudofossils. Was the origin of these objects biotic excretion or abiotic extrusion? Past evidence is not sufficient for resolving this issue. New information strengthens the abiotic interpretation, but these ferruginous specimens remain as a geologic enigma.

1. Introduction

The nature of ferruginous objects that have shapes that resemble animal excrements has long been debated, with proponents deeply divided as to biologic and abogenic origins. The goal of this paper is not to resolve this controversy but to instead provide an overview of occurrences and to describe various types of evidence that can be used to make valid interpretations as a substitute of the all-too-common “I know poop when I see it” generalization. The main focus is on specimens from the Miocene Wilkes Formation at Salmon Creek, Lewis County, southwest Washington State, USA. This site provides an opportunity to study abundant ferruginous “coprolites” at a location where the geologic setting is well studied.

Coprolites are a common trace fossil, first recognized in the late seventeenth century [1,2], and described in detail by Buckland, who introduced the “coprolite” name for fossil feces [3]. In 1968, Häntzschel et al. [4] published an extensive annotated bibliography of coprolites. Updated information can be found in [5]. The terminology has been refined, using “bromalite” to describe trace fossils that are related to food processing. This category includes objects having diverse morphologies. Examples include the regurgitation of undigested remains produced by raptors, or stones swallowed by herbivorous dinosaurs (gastroliths). Objects traditionally described as coprolites (fossil feces) are now known to include lithified intestinal contents (cololites). Terminology is explained in detail by [5,6,7].

The study of invertebrate trace fossils has long been dominated by the belief that ichnotaxonomic names should be based only on morphology with no consideration of the producer of the trace fossils. This ichnotaxonomy is in marked contrast to the traditions of vertebrate ichnologists, who commonly attempt to identify the life forms that produced footprints. For bromalites, these interpretations have included invertebrate as well as vertebrate producers (e.g., [5,8]).

Feces originate as a process of excretion of material that remained undigested during passage through the intestinal tract. The fossilization of feces is strongly related to the elemental composition. The preservation of carnivore coprolites is favored by included skeletal fragments that are a source of calcium and phosphorous (e.g., apatite); more than 90% of coprolites have been observed to contain calcium phosphate as the primary mineral component [4]. Coprolites from herbivorous animals are relatively rare, because these excretions typically contain vegetative remains that are rapidly composed. The widespread use of bovid manure as an agricultural fertilizer is evidence of this phenomenon. The occurrence of herbivore excrements is largely limited to Quaternary deposits (e.g., [9,10,11]).

The most enigmatic variety of purported bromalites are those that consist almost entirely of iron minerals, which range from nearly pure iron carbonate (siderite), to iron oxides and hydroxides (e.g., goethite). Some localities contain specimens that are mixtures of these minerals. The identification of these ferruginous objects as coprolites or cololites was originally based on subjective observations, as typified in the declaration made by Amstutz [12] in regard to Salmon Creek specimens: “Many specimens are clearly dung-shaped and leave no doubt as to their origin” [12] (p. 502). The problem with this interpretation is that the ferruginous objects at all major localities lack features that are common for excrements or intestinal casts. These characteristics include a limited range of sizes and shapes, the presence of inclusions of undigested matter (e.g., hairs, bone fragments, or plant fibers), and the stratigraphic coexistence of fossil evidence of possible coprolite producers. Pointed ends and longitudinal striations are common in coprolite from terrestrial producers.

In contrast, ferruginous specimens typically have the following characteristics:

• Strata that contain ferruginous “coprolites” preserve no vertebrate fossils. Microfossils in the sediment may include pollen and diatoms.
• The ferruginous masses contain no inclusions of undigested material (e.g., skeletal fragments, fish scales, or vegetative fibers).
• The paucity of phosphorus precludes the possibility of excrement by a carnivore.
• Ferruginous masses occur in a wide range of sizes and shapes, and “coprolites” include those that can be found in strata that contains non-extrusive shapes.
• Unbiased collection shows that a large percentage of specimens are very small.
• An abundance of unmineralized wood occurs in direct association with ferruginous “coprolites”. At Salmon Creek, Washington, some coprolite-like specimens are attached to or found within ancient wood.

These phenomena are discussed in detail later in this report.

2. Global Occurrences

Table 1. Reported occurrences of ferruginous “coprolites”.

Age	Location	Formation	References
Quaternary	Five Docks, New South Wales, Australia	Clay beds	[13]
Late Miocene	Lewis County, Washington State, USA	Wilkes Formation	[12,14,15]
Miocene	Madagascar	Not reported	This report
Miocene	Nundle, New South Wales, Australia	Not reported	[16]
Miocene	Poland	Turów Lignite	[17]
Eocene	North Dakota, USA	Fort Union Formation/Golden Valley Formation	[18,19]
Upper Cretaceous	Saskatchewan, Canada	Whitemud Formation	[20,21,22,23,24,25,26,27]
Upper Cretaceous	Alberta, Canada	Oldman Formation	[28,29]
Permian	Arkhangel’sk, Russia	Not reported	[30]
Upper Permian	China	Not reported	[19]

lists known occurrences of ferruginous “coprolite” specimens. Three international localities have yielded thousands of specimens: the Miocene Wilkes formation in southwest Washington State, unmanned Miocene beds in Andilamena District, Madagascar, and the Upper Cretaceous Whitemud Formation of Saskatchewan, Canada. This report includes data from all three localities, but the main focus is on specimens from late Miocene Wilkes Formation beds exposed along Salmon Creek in Lewis County, Washington. We include a brief overview of other localities. These include three sites where coprolite-shaped ferruginous concretions occur: The Morrison Formation in south-central Utah, USA; White Desert, Egypt; and Western Australia.

Ferruginous masses in Wilkes Formation strata have diverse sizes and shapes (Figure 1). Variations in size and shape are also characteristic of specimens from Saskatchewan, Canada and Madagascar, two other localities where ferruginous masses occur in enormous abundances.

Eleven specimens were collected in 2011 by University of Regina professor Martin Beech from a clay pit ~4 km northeast of Willow, Saskatchewan, supplemented by 12 specimens collected in 1996 by Brian McManus (Figure 2). These specimens likely have some degree of collecting bias, and the release of specimens by weathering and their subsequent fluvial transport reduces the presence of small specimens. Despite these limitations, the specimens display a wide range of morphologies.

Madagascar specimens (Figure 3) were purchased online in 2025 from vendors in China. The specimens are the products of biased collecting, because residents of villages near Marovato in the Andilamena District, Madagascar gather the samples as a source of income. Market values dictate the gathering of only specimens that look like excrements. Madagascar specimens are typically advertised as dinosaur coprolites, with a purported Jurassic age, but the true age is likely Miocene based on regional geology.

Goethite “coprolites” from the Morrison Formation, Utah, USA, are represented by 48 specimens that were acquired from the personal collection of Dylan Hart, an amateur fossil enthusiast. Goethite pseudomorphs of marcasite and pyrite from White Desert, Egypt were purchased online in 2025 from several vendors in China (Figure 4).

Sinuous goethite nodules that have been eroded from Upper Jurassic Morrison Formation strata in south-central Utah, USA have long been gathered by amateur collectors, who regard the objects to be coprolites. These forms have not received scientific scrutiny. The Morrison Formation forms extensive beds in Wyoming and Colorado, but outcrop zones extend into adjacent states. The strata typically consist of continental deposits of clastic fluvial sediments that interfinger with aeolian sandstone. For ichnologists, the Morrison Formation is well known as a source of purported dinosaur coprolites, but collecting of these specimens has involved indiscriminate gathering of siliceous materials [31].

Ferruginous nodules at other locations include sinuous forms that are reminiscent of excrement. Examples include specimens that occur abundantly on the surface in desert sands in the White Desert north of Farafara Oasis, in the New Valley Governorate, Egypt (Figure 5). First described in 1898 [32], the specimens were identified as being pseudomorphs after marcasite, but the despite later study, the origin of the diverse shapes has not been elucidated [33]. Large quantities of these Egyptian nodules are presently being exported to vendors in China, who market them on the internet as “prophecy stones” that purportedly have metaphysical virtues. These specimens are typically labeled as goethite and hematite, but XRD reveals goethite as the only iron mineral. Crystal form suggests that the pseudomorphs replaced both pyrite and marcasite (Figure 5).

Goethite pseudomorphs after marcasite occur in the Gascoyne River area, Western Australia [34]. Some elongate specimens resemble coprolites (Figure 5R–U), but polycrystalline masses from both locations clearly show the pseudomorph origin (Figure 5O–Q). The Egypt and Australia specimens illustrate the diversity of sinous specimens that range from objects that have been interpreted to be coprolites or cololites to specimens of similar composition that are clearly of abiogenic origin. Boundaries between the two groups are not well defined.

3. Materials and Methods

Thin sections and polished cross-sections were prepared using facilities at the Geology Department, Western Washington University, Bellingham, WA 98225, USA. Photomicrographs were made using a Zeiss petrographic monocular microscope (Carl Zeiss Microscopy, White Plaiuns, NY, USA) equipped with a Hayear 500B 5-megapixel CMOS microscope camera (Shenzen Hayear Electronics, Ltd., Shenzen, China). SEM images came from a Tescan Vega III scanning electron microscope (Tescan, Brno, Czech Republic) at the W.W.U. University Instrument Center. Specimens were mounted on 1 cm diameter aluminum stubs using expoxy adhesive and sputter coated with Pd to provide electrical conductivity. Images were made with a beam energy of 10 KV. X-ray diffraction patterns were obtained from packed powders using a Rigaku Miniflex 6G diffractometer controlled by Smartlab Studio II software (Rigaku Corp., Tokyo, Japan), located at the W.W.U. Advanced Materials Science and Engineering Center.

4. Salmon Creek, Washington

Wilkes Formation beds at Salmon Creek, Washington State, USA offer a unique opportunity for studying ferruginous “bromalites” because it is the only occurrence where specimens can be collected in situ from beds that have well-documented stratigraphy and precisely known radiometric age. Paleoenvironmental and paleoclimatic conditions have been analyzed using palynology and sedimentology. In this report we supplement that information with data from optical mineralogy, SEM, and X-ray diffraction analysis.

4.1. Specimen Collecting

Research material consists of approximately 1000 specimens that were collected from seven Salmon Creek localities during visits between 1980 and 2025 (Figure 6,

Table 2. Salmon Creek localities.

Locality	Informal Name	Latitude	Longitude	Elevation (Meters)
WWU-82177	HB upstream	46°23′27′′ N	122°47′45′′ W	48
WWU 1571	Horseshoe Bend	46°23′26′′ N	122°47′40′′ W	47
WWU-82172	Red Gate	46°23′24′′ N	122°47′29′′ W	47
WWU-6239	Cougar Creek confluence	46°23′36′′ N	122°47′38′′ W	48
WWU-82174	Tooley Road Bridge	46°24′46′′ N	122°47′38′′ W	40
WWU-82176	Upstream from Jackson Bridge	46°25′23′′ N	120°48′54′′ W	40
WWU-82175	Jackson Highway Bridge	46°25′23′′ N	122°50′24′′ W	33

).

Early studies of specimens from Salmon Creek were based on specimens that were collected from the stream bed upstream from the Jackson Highway Bridge (WWU-82175, 82176; Figure 7A), a locality where specimens typically contain siderite as the major component (Figure 8A–E). In marked contrast, specimens of similar morphology collected in situ from upstream sites in the Wilkes Hills (WWU-1571, Figure 7B) are mostly composed of oxidized iron minerals, e.g., limonite, goethite (Figure 8F–L), with siderite being present in some specimens.

Private property owners have limited access to Salmon Creek in the Jackson Highway Bridge area, but vast numbers of specimens have been commercially collected from the middle section of the stream, near the Tooley Road Bridge (WWU-8217). Specimens from this locality have commonly experienced stream transport, but the lack of erosion suggests that strata exposed on nearby stream banks were a local source. The search for specimens in the streambed near the Tooley Road Bridge has been inhibited by changes in the stream pattern, where shallow gravel bars have been replaced by deep pools (Figure 9). Also, collecting opportunities are greatly affected by seasonal changes in stream flow, with stream bank access being best during the low water levels that occur in late summer.

The extent of commercial collecting at Salmon is evidenced by a conversation that I had with a collector during a field visit in 1994. This person said that three commercial collectors were gathering specimens to sell on a wholesale basis. This particular collector had sold ~2.270 kg (5000 lbs) of Salmon Creek “coprolites”, most of them purchased by a dealer in Arizona, USA. These specimens show strong collecting bias because of marketing pressures that favor dung-shaped specimens (Figure 10).

At upstream localities in the Wilkes Hills, commercial forest lands are largely under ownership of the Weyerhauser Corporation (https://www.weyerhaeuser.com), and public entry requires payment of annual access fees. The company has been cooperative about allowing property access to researchers, but recreational visitors are met with prominent signs that forbid rock collecting.

4.2. Geologic Setting

The Wilkes Formation comprises a sequence of fluvial and lacustrine sediments that are located in a region of southwest Washington State where Cenozoic sediments are interbedded with extrusive volcanic rocks [35]. Local dip angles seldom exceed a few degrees.

The primary components are semi-consolidated tuffaceous siltstone and sandstone. Unmineralized fossil wood is very abundant, particularly in the lower part of the stratigrapic section, consisting of upright stems and trunks, and large transported fragments. The late Miocene age of the Wilkes Formation was originally estimated from fossil leaves collected from strata in the lower part of the formation [36]. More recently, an ash bed at a Salmon Creek site in the Wilkes Hills yielded a ⁴⁰AR-³⁹AR age of 6.13 ± 0.08 MA [37].

The stratigraphic positions of beds that contain ferruginous masses are not known with certainty, but Roberts [35] inferred them to be in the lower part of the Formation. Salmon Creek has a meandering channel pattern, with a gentle gradient, and the near-horizontality of the Wilkes Formation strata suggests the possibility that all of the collecting sites are in a correlative stratum. At sites along Salmon Creek, the “coprolite”-bearing bed is typically the highest exposed unit, where the Miocene sediment has a sharp contact with a surface layer of Quaternary alluvium (Figure 11).

4.3. Stratigraphy

The Wilkes Formation has an estimated total thickness of 32 m (760 ft), based on incomplete stratigraphic sections that were mostly measured at road cuts along State Highway 1Q. The various lithologic components interfinger and have lateral transitions, so no individual section is typical of the whole formation. The 32 m composite section reported by Roberts [35] did not reveal “coprolite”-bearing strata, but this horizon is intermittently exposed along the course of Salmon Creek. Gentle topography and heavy streamside vegetation limit sedimentary visibility. In 2007, Yancey and Mustoe measured a section at an upstream locality in the Wilkes Hills where 19 m of strata are exposed [37,38].

This section (Figure 12) includes beds that contain abundant “coprolites”.

The Wilkes Formation sediments were deposited during a time of major geologic change in the Puget Lowlands that lie west of the Washington Cascades. In the middle Miocene, regional volcanic activity transitioned from the flows of basaltic lava of the Columbia Plateau to the eruption of andesitic and dacitic volcanoes that were an early part of the “ring of fire” that parallels the continental margins of North and South America. The sedimentology of the Wilkes Formation is evidence of a cycle of lahar flows associated with ancient eruptions. These cataclysmic events episodically inundated a low-lying structural basin; volcanic mudflows (lahars) disrupted drainage patterns, locally ponding water to produce lakes and swamps where woody peat accumulated. Continuing rise of the Cascades uplifted and gently deformed the Wilkes Formation sediments, ending sediment accumulation [35,37].

Wilkes Formation strata record repeated lahar flows. Each depositional cycle is characterized by a thick layer of clastic sediment that fine upward from siltstone to silty clay, evidence of gravitational settling. Each cycle terminates with a stratum rich in wood fragments in horizontal orientation. This material originated by the settling of water-logged wood debris that was fluvially transported following an eruption that decimated an upstream forest (Figure 13A,C). Similar lahar deposits were produced in modern eruptions (Figure 13B,D)

Ancient wood is abundant in Wilkes Formation sediments, comprising upright vegetation buried during the influx of lahar sediments, large fluvially transported trunk fragments, and the wood mats described above. Wilkes Formation wood is not mineralized, but specimens commonly show shrinkage, shape distortion, and localized carbonization (Figure 14).

4.4. Palynology

Wilkes Formation strata contain well-preserved pollen and spores from diverse plant communities (Figure 15). These microfossils trace ecological succession that accompanied episodic lahar events. Cypress swamps were a common environment, interspersed with episodes of water retention in the form of shallow lakes. Cataclysmic forest destruction was followed by botanical succession in the form of fern glades and deciduous forest, ultimately transitioning back to cypress swamps, with water that was inhabited by algae and aquatic plants [39].

4.5. Evidence from Upstream Locations

Ferruginous masses occur at several stratigraphic levels in the beds that are exposed along Salmon Creek and its tributaries in the Wilkes Hills, but at each locality coprolite-shaped forms are limited to a single stratum. Other clay-rich beds contain objects of similar composition but having different morphologies. Roberts [34] observed an abundance of disc-shaped and botryoidal forms. At the WWU-8217 locality, an argillaceous siltstone layer contains abundant in situ concretionary masses (Figure 16).

Site WWU-1571 is a prolific source of “coprolite” specimens, which occur at the highest stratigraphic horizon. Because of the steep topography, access to this bed is difficult. Collecting opportunities are facilitated when large blocks fall to streamside level (Figure 17).

4.6. Morphologies of Specimens

In 2025, I collected approximately 200 specimens from recently fallen blocks at WWU-1571 (Figure 17). Every observed specimen was collected, without preference for size or shape. A plot of specimen weight (Figure 18) reveals the abundance of small specimens very different from the biased specimens that favor “coprolites” (e.g., Figure 9). Regardless of size, specimens that have extrusive shapes have reasonably uniform length/width ratios (Figure 19). The graphical data is consistent with observations of very small specimens, which are similar in morphology to the classic larger “coprolite” specimens (Figure 20). This similarity is probably due to the extrusion of material that had fairly uniform viscosity.

The graphed data used measurements only from specimens that had sinuous form. Other specimens in the same stratum have botryoidal concretionary shapes (Figure 21).

The sediment that contains “coprolites” also contains large masses of ferruginous material, where clastic sediment contains abundant ferruginous cement (Figure 22). These specimens are evidence that the precipitation of iron minerals was not restricted to replacement of organic matter.

4.7. Transect Study

By the mid 1990s, recreational and commercial collectors discovered the upstream beds along Salmon Creek, and digging began at the highest Wilkes Formation stratum exposed at site WWU-1571 (Figure 23A). In 2009, I excavated this site to create a measured 3-D transect where the taphonomy of in situ specimens could be studied. The dimensions of the face of the excavation were 0.75 × 0.5 m, explored to a depth of 10 cm (Figure 23B). The dense cohesiveness of the damp, clay-rich sediment inhibited unbiased collecting, because very small specimens were difficult to see. However, the distribution and orientation of larger specimens could be determined, and unfossilized wood was easily recognizable (Figure 23F).

4.8. Association of “Bromalites” with Buried Wood

A striking characteristic of Salmon Creek in situ localities is the association of ferruginous “coprolites” with buried wood, a phenomenon that is present at downstream sites (e.g., beds upstream from Jackson Highway Bridge, and upstream sights in the Wilkes Hills). Roberts [35] commented that “Some coprolitic forms were found surrounded by carbonaceous material”, a fact suggesting that the shape of some concretions resulted from being squeezed through roots and stems of aquatic plants (p. 36).

Wood may be in the form of upright stems (Figure 23F and Figure 24) or larger wood fragments (Figure 24). Modes of occurrence include sideritic extrusions that occur within unmineralized wood (Figure 25) and ferruginous masses that surround a buried stem (Figure 26). A few specimens have the external shape of wood, where ferruginous material has formed as a cast that preserves the external shape of a stem (Figure 27). An excremental origin is unlikely for these wood-associated specimens.

5. Mineralogy

The mineralogy and petrogensis of ferruginous “bromalites” have been subject to diverse interpretations. The Wilkes Formation specimens have long been considered to be composed primarily of siderite, with oxidized iron minerals commonly present as a thin outer coating that represents diagenetic alteration. This composition was described by Amstutz [12], who reported a polished thin section that showed marked zoning, with a rind of about 80% siderite grains in a matrix of limonite, with the central area composed of 90%–95% siderite. Roberts [35] likewise observed coprolite-like concretions that consisted of granular siderite.

During my investigation, specimens from various Salmon Creek localities were found to have diverse compositions and textures (Figure 28). The mineralogy was studied using a variety of methods that included thin section petrography, X-ray diffraction, and SEM/EDS.

Specimens collected from upstream sites (e.g., WWU-1571) are commonly composed of iron oxides or hydroxides (Figure 29). One possibility is that these specimens underwent diagentic oxidation that caused the original siderite to have been replaced by limonite, goethite, or hematite, or mixtures of these minerals.

Specimens that have simple extrusive shapes may show a distinctive peripheral zone (Figure 30). In contrast, specimens that show botryoidal overgrowths commonly have layers that follow the concretionary forms (Figure 31).

5.1. Thin Section Petrography

Figure 32 shows thin sections that have the type of mineralogy that was described by Amstutz [12].

5.2. Scanning Electron Microscopy (SEM)

SEM is an important tool for studying geologic specimens, particularly when the electron beam imaging is combined with an X-ray spectrometer that allows semiquantitative analysis of elements in the sample. The inability of XRF spectrometry to detect hydrogen limits the ability to discriminate between oxide and hydroxide minerals. Likewise, oxidation states of iron cannot be ascertained, and software-based quantification will arbitrarily calculate total iron as FeO or Fe₂O₃. These analytical limitations can be reduced by combining SEM/EDS data with mineral compositions as determined by XRD. A difficulty is that SEM/EDS can provide elemental abundance information for a specimen area that is only a few microns in diameter, but XRD data come from a powdered sample of much larger volume.

For Salmon Creek specimens, SEM images allow detailed views of the morphologies of the various minerals that comprise ferruginous “coprolites”. Figure 33 shows the crystal architecture of materials in Salmon Creek specimens. Mineral identifications are based on XRD patterns and SEM/EDS spectra.

The great depth of field of SEM images provides viewing of textural relationships (Figure 34). For Salmon Creek specimens, important features include the common existence of porosities within the interior zone of ferruginous “coprolites”, which may be in the form of crystal-lined vugs (Figure 34A,B). In some specimens, the peripheral zone contains diatoms frustules that are incorporated within the iron-rich material.

Freshly excavated ferruginous specimens typically have thin orange surface coatings (Figure 35). XRD shows the composition of this material to be a mixture of lepidocrocite (FeOOH) and quartz. In sedimentary environments, lepidocrocite results from slow oxidation of Fe² under anaerobic conditions in a temperate climate [39]. For Salmon Creek specimens, the orange coating thus represents “rusting” of the ferruginous mass, incorporating clay and silt from the adjacent claystone. The unconsolidated nature of this layer is evidenced by the ability of ultrasonic cleaning to remove this material. SEM images (Figure 35) reveal the physical differences between the surface coatings and the underlying material.

5.3. Density

Densities of ferruginous “coprolite” specimens from four localities were measured using hydrostatic weighing, where the dry weight of the specimen is divided by the weight of the specimen when it is suspended in water (

Table 3. Densities of ferruginous “coprolites”.

Wilkes Fm Washington State, USA		Whitemud Fm. Saskatchewan, Canada	Morisson Fm Utah, USA	Madagascar
WWU-82175	WWU-1571
3.63	3.07	2.92	2.99	2.92
3.66	2.68	2.90	3.15	2.81
	3.04	2.95	3.12	3.51
	2.85	2.85	3.13	2.86
	2.92	2.78	2.86	3.36
	2.53	2.93	3.11	2.80
	2.87	2.93	2.86	3.52
Mean value
4.31	2.85	2.89	3.05	3.11
Reference data
Siderite	3.96
Hematite	Up to 5.26
Goethite	3.3–4.3
Ferrihydrite	3.8
Limonite group	2.7–4.3

). The most dense specimens (Figure 28A,B) were Salmon Creek site WWU-82175 extrusions that contain siderite as the primary constituent, but the values for both specimens are somewhat lower than the density of pure siderite. These differences perhaps represent lower densities of sedimentary siderite compared to hydrothermal or metamorphic siderite. Other specimens from all localities have densities that are below the values typical of well-crystalline iron minerals, which suggests that these materials contain varying amounts of low-density amorphous iron hydroxides of the limonite group.

5.4. X-Ray Diffraction

X-ray diffraction patterns are useful for recognizing the presence of crystalline materials, but the inability of XRD to detect amorphous phases is a limitation when the method is used to characterize the mineral compositions of ferruginous specimens. Salmon Creek specimens were observed to contain goethite as a major component, with siderite commonly being present in varying amounts (Figure 36). No other iron minerals were detected.

As discussed later in this report, X-ray diffraction data from Salmon Creek specimens contradict depositional models that presume ferrihydrite is the primary precipitate and that exterior zones incorporate clay and silt from the adjacent clastic sediment to produce a hardened layer.

5.5. Siderite Geochemistry

The petrogensis of siderite has received considerable study, though much of the data comes from experiments that investigate higher temperatures and pressures than are typical of sedimentary environments. In general, siderite is a primary product in reducing environments where there is a source of dissolved iron and an availability of carbonate, which may come from decomposition of organic matter. Under oxidizing conditions, siderite is likely to ultimately alter to hematite, with ferric hydroxides or goethite as intermediaries [40,41,42,43].

The stability field for siderite is relatively narrow, requiring only mildly reducing anaerobic conditions at acidic or neutral pH, a typical range for sediments in lacustrine or palludal environments (Figure 37). If sulfur or phosphorus is present, the range for precipitation of siderite is narrowed because of the possible formation of Fe sulfides (pyrite or marcasite) or Fe phosphate (vivianite). The geochemical data supports the hypothesis that the initial mineralization at Salmon Creek was in the form of siderite, in contrast to the interpretation that siderite was a late-stage precipitate [25,26].

5.6. Oxidation Processes

Seilacher et al. [19] suggested that fecal material was affected by “roll front” action of migrating groundwater that produced acidic conditions that dissolved calcareous and phosphatic materials, accompanied by precipitation of siderite. There are several problems with this hypothesis. One is that studies of roll-front processes have largely focused on the formation of uranium ores, particularly in the western United States. In these deposits, the flow of groundwater typically occurs along paths of high permeability in coarse-grained classic sediments (e.g., sandstone that was deposited in paleochannels). In roll front deposits, the precipitation of uranium and vanadium minerals has commonly been concentrated in buried logs, where degradation of the plant tissues produced local reduction gradients [44]. These conditions are very different from the occurrences of ferruginous “bromalites”, which are invariably found in clay/siltstone beds that have low groundwater permeability. Also, the same beds that contain abundant ferruginous masses also preserve large quantities of unmineralized wood. It is unlikely that a roll front process could cause fecal masses to be entirely replaced by iron minerals, accompanied by destruction of all animal fossils, but with no alteration of the associated botanical remains.

The replacement of fecal material by siderite is a known process, as evidenced by the discovery of ferruginous coprolites at a Miocene locality in Poland [17]. However, the occurrence of sideritic specimens at this locality is very different from the deposits In Saskatchewan, Washington State, and Madagascar. The excrement-shaped masses at the Turów lignite mine in southwest Poland are relatively scarce, and they have relatively simple shapes and rather uniform sizes (Figure 38). The specimens preserve inclusions that include hair-like fibers and carbonaceous materials. Age-correlative strata at other sites contain rich faunal assemblages, with aquatic tetrapods that include frogs, salamanders, crocodiles, turtles, as well as lizards, and snakes. The authors of the study compared the coprolites to excrements from a diverse variety of zoo animals, and they noted the strongest correspondence of the fossils to feces from tortoises and snakes. This interpretation is supported by the discovery of a fossil tortoise shell fragment in the same stratum that contained the coprolites (Figure 38E).

5.7. Microbial Processes

The early recognition that Salmon Creek specimens could primarily consist of siderite led to a common belief that ferruginous “coprolites” at other localities had a similar mineral origin. For example, for the Whitemud Formation, Broughton et al. [22] speculated that original fecal material was replaced by siderite and pyrite, which was subsequently altered to limonite, goethite, and lepidocrocite. The later discovery of specimens that were interpreted to be intestinal casts of herbivorous tetrapods was accompanied by the continued belief that the specimens originated as siderite replacement of excrement [24].

That mineralogic interpretation was later rescinded, replaced by the interpretation that coprolites and cololites were preserved as 3-dimensional casts as a result of the formation of a hard crust produced by mat-forming microbes. The hypothesis [25,26] is that intestinal bacteria flourished in the fecal material to form a peripheral microbial mat that caused ferrihydrate precipitation. This iron hydroxide rapidly transformed to goethite, which in combination with silt grains produced a layer of cemented sediment that acted as a mold for the external shape of the organic matter. Continued microbial precipitation of ferrihydrate (with goethite as an alteration product) led to concentric expansion to produce a 3-D cast of the original feces. In this model, siderite may occur as a secondary stage of mineralization.

This microbial mat hypothesis has a number of deficiencies. A fundamental weakness is the assumption that members of the gut biome can produce mats that precipitate iron minerals. Although intestinal microbes can include archaea, fungi, and viruses, bacteria make up most of the microflora in the colon, where they facilitate digestion for the host animal by expanding nutrient sources and production of essential vitamins. For example, in herbivores the ability to digest cellulose is possible because of gut microbes [45]. Gut microbes are heterotrophic (saprophytic), depending on the degradation of organic matter as a source of metabolic energy. In contrast, iron bacteria are chemotrophic, deriving their energy from chemical reactions, or mixotropic, where a bacterium can metabolize organic matter but flourishes by oxidation of reduced iron. The best-known iron-oxidizing bacteria are species within the genus Leptothrix (Class Gammaproteobacteria, formerly known as Betaprotobacteria). A distinguishing characteristic of Leptothrix is the production of iron oxide-rich deposits that develop around sheaths that are devoid of living cells. These long filaments originate as chains of individual cells that are surrounded by a gelatinous sheath. Leptothrix iron bacteria occur in neutral or slightly acidic aerobic fresh-water environments that provide a source of dissolved ferrous iron [46,47,48].

Iron oxidizing bacteria in the family Gallionellaceae include members of the genera Gallionella, Sideroxydans, Ferriphaselus, Sideroxyarcus, and Ferrigenium [49,50,51]. These fresh-water bacteria have diverse environmental ranges that include creeks, wet sediment, peat, permafrost, deep subsurface aquifers, and municipal waterworks. Species of this family tend to be mixotropic, able to survive by metabolizing organic compounds, but flourishing when there is an opportunity to oxidize dissolved iron. Sideroxydans can metabolize Fe from a mineral substrate, e.g., magnetite [47], an ability that is perhaps be present in other members of the Gallionellaceae.

The characteristics of fresh-water iron biology appear to preclude their presence as gut bacteria. Other factors are contrary to the hypothesis that bacterial mats might have produced cemented peripheral zones that caused fecal material to be entombed within a protective coating. In particular, the iron minerals precipitated by aquatic iron bacteria are soft, gelatinous masses, not rigid mats. This phenomenon is evidenced in ferruginous “coprolites”, which commonly have extensive synaresis cracking on their outer surfaces (Figure 39). Synaresis occurs when water is expelled from masses of soft, hydrous material. In the Salmon Creek specimens, synaresis cracks are limited to the peripheral zones, suggesting that these layers originated as the gelatinous ferric precipitates typical of extant fresh-water iron bacteria. Second, their ability to metabolize dissolved ferrous iron makes them unlikely to have been the agents for initial biomineralization. A more likely scenario is that these bacteria inhabited the surface of objects that had already acquired a ferruginous composition. Under those circumstances, the outer zones that can be observed in some sawn specimens probably represent oxidative alteration, not primary precipitation (Figure 30). This process is evidenced by specimens that have botryoidal encrustations, where the alteration layers developed on the concretionary surfaces (Figure 31).

The development of an oxidized outer zone appears to commonly develop from the oxidation of the interior to form a porous zone that can bond to mineral clasts in the adjacent sedment (Figure 40). This is contrary to the hypothesis that an iron oxide layer forms as an initial microbial mat, with minerals growing inward from that outer layer [25,26].

The inference that the interior regions of the ferruginous masses were produced by the inward growth microbial mats [25,26] was based on a small number of micro-CT scans, including data from only two specimens from Salmon Creek. I prefer the direct viewing that results from polished cross-sections, none of which show evidence of microbial mat fabrics. The absence of these fabrics was suggested to have been the result of siderite remineralization, an interpretation that we believe is a generalization that was made to support a pre-determined conclusion.

Another problem with the hypothesis of microbial biomineralization is that the sequence is inferred to have begun with precipitation of ferrihydrite, with goethite and hematite as subsequent alteration products [25,26]. However, XRD data, SEM images, and optical petrography of thin sections all show an apparent absence of ferrihydrate and hematite in Salmon Creek specimens. The most common paragentic process appears to be the abiogenic oxidation of siderite to produce goethite.

A fundamental factor for biogenic precipitation of iron minerals is that microbes that metabolize dissolved iron as a source of metallic activity are most likely to flourish in environments that contain iron-rich sources. An example is the common occurrence of Leptothrix coatings on the inner surfaces of steel water pipes. For ferruginous “coprolites”, bacterial iron precipitation is more likely to occur as a late-stage cause of surface alteration, not the primary phase of mineralization, as proposed by Broughton [25,26].

6. Discussion

The study of ferruginous specimens has been clouded by investigations that have had a variety of weaknesses.

6.1. Uncertainties in Earlier Research

As discussed earlier in this report, speculations on the origin of ferruginous “bromalites” have often been based on specimens that were subject to collecting bias. An associated problem has been interpretations that were based on minimal data. For example, the first description of Salmon Creek ferruginous extrusions as cololites [19] was based on two specimens that were purchased at a Connecticut rock show. The only other evidence used to support that cololite interpretation was a single sinuous specimen whose location was merely described as being from China and a reference to anecdotal observations of a North Dakota Miocene occurrence, which was illustrated by line. drawings of two specimens (Figure 41).

6.2. Problematic Ichnotaxonomy

The establishment of a new ichnotaxon, Hirabromus seilacheri Hunt et al., was based on the small number of ferruginous “bromalites” in the collection of the United States National Museum of Natural History (Smithsonian Institution) [52]. The proposed ichnogenus comprises a single ichnospecies. The diagnosis was described as follows: “Bromalite that differs from other ichnogenera in consisting of an elongate, convoluted, longitudinally-striated cylinder with tapering terminations composed of siderite”.

In the text, the holotype is described as a specimen from the Paleocene Fort Union Formation near Rhame, North Dakota, USNM catalog number of 313350. However, an accompanying illustration identifies the holotype as Salmon Creek, Washington specimen SNM 313,348 (Figure 42). Another concern is that the Hirobromus diagnosis states that the ichnogenus comprises specimens composed of siderite. The use of mineral composition as a diagnostic ichnotaxonomic characterstic is a novel concept, questionable in this instance because referred specimens include ferruginous examples from localities where siderite is not known to be present. Also, the absence of Saskatchewan specimens in the USNM collection precludes recognition of specimens that may be the closest morphotypes to the Hirobromus holotype. However, a limitation for applying a binomial system of nomenclature to ferruginous bromalites is that the observed shapes are commonly affected by nonbiologic processes, such as synaresis cracks and concretionary overgrowths [20].

6.3. Use of Problematic Specimens

An example of interpretation based on unverifiable data was the use of a single 1.04 m-long “cololite” to support the hypothetical origin of ferruginous specimens from the Wilkes Formation in Washington State [25,52]. In 2014, this specimen was sold to an anonymous buyer at a private auction in Beverly Hills, California. The present location of the specimen is unknown. The specimen had been collected somewhere in the Salmon Creek area by an unidentified property owner who did not disclose the collecting locality. No stratigraphic or sedimentologic information was recorded.

6.4. Subjective Data Presentation

Broughton [24] described four presumed intestinal casts that he excavated from a Whitemud Formation locality. The length of the largest specimen was reported to be 75 cm plus an additional 30 cm partially eroded segment. Two other specimens had reported lengths of 20 cm each, with a fourth specimen of 50 cm length. In a later paper, the same four specimens were described as having lengths of 1.0–1.5 m [53].

That author [53] described the occurrence of three major deposits in northwestern North America, “each with hundreds of thousands of unusually shaped sideritic specimens”. Those descriptions may be plausible estimates for the Whitemud Formation in Saskatchewan and the Wilkes Formation in Washington State, but high abundance is not verified for the Golden Valley Formation of North Dakota. The cited reference for the purported abundance of coprolites and cololites at the North Dakota site (Seilacher et al., 2001) [19] was anecdotal observations by one of the paper’s coauthors, who provided sketches of only two specimens. Their stratigraphic position was illustrated with a stratigraphic illustration that was cited as coming from Hickey, 1977 [54]. However, the illustration does not appear in that publication. Moreover, Hickey observed no coprolite-like specimens in the Golden Valley Formation. An earlier investigation [55] noted that some Golden Valley Formation beds contained abundant coprolites, but these are simple ovoid shapes that have non-ferruginous compositions. The nearest support for the presence of ferruginous North Dakota “coprolites” comes from Brown [18], who illustrated four specimens that he collected from the Fort Union Formation, which underlies the Golden Valley Formation.

6.5. Fanciful Early Interpretations

Major [56] believed the ferruginous shapes at Salmon Creek were pseudomorophs of bryozoans. Dake [57] proposed that the objects formed in caves as speleothems that were contorted by shifting air currents. These fanciful explanations were quickly discounted, replaced by the suggestion that the objects had an excremental origin [12].

6.6. Interpretations Based on Pre-Conceived Conclusions

The belief that ferruginous specimens can only have had a digestive origin has resulted in many subjective evaluations.

A particular challenge for advocates of the bromalite hypothesis has been to provide an explanation for the absence of undigested remains within the specimens. In the case of ferruginous specimens from Permian strata at Arkhangel’sk, Russia, the absence of plant or animal remains in thin section views resulted in the assertion that the masses were excrements from reptiles that ate insect larvae, slugs, and other soft-bodied organisms, accompanied by ingestion of sedimentary ooze [30]. The hypothesis accounted for the production of fecal matter that contained clastic mineral grains but no organic residues. However, the explanation failed to explain why no skeletal remains of ancient animals were preserved.

In the case of the Upper Cretaceous Whitemud Formation in Saskatchewan, Canada, an early hypothesis was that the ferruginous masses were replacements for feces of sturgeon or bowfin, two types of fish that were known to exist in the Mesozoic Era, with cartilaginous architecture that could explain the absence of skeletal remains [22]. This interpretation was changed in 1981, following the discovery of several specimens that were inferred to be intestinal casts [23,24]. These were claimed to be evidence of a teleost (bony) fish or an aquatic tetrapod, with a suggestion that the coprolites and oncolites could have been made by members of a single genus, and perhaps a single species [24]. As in the Russian example, the absence of skeletal remains was unexplained.

The presence of extruded specimens spanning a wide size range has also been a challenge for investigators who have committed to an excremental origin for the ferruginous extrusions. Schmitz and Binda [27] asserted that the Whitemud Formation clay-rich beds that contained “coprolites” had been produced by subaerial rather than subaqueous deposition. These authors suggested that the diversity of specimen sizes were evidence that the deposition occurred on an episodically inundated flood plain, creating seasonal water holes that attracted a wide variety of coprolite-producing animals, possibly including crocodiles, turtles, flying reptiles, dinosaurs, and small mammals. They grouped several hundred ferruginous specimens into nine morphotypes based on coiling, tapering, elongation, and surface texture. Rough-textured specimens were inferred to have been excreted by herbivores, with smooth-textured specimens produced by carnivores. Once again, the production of a multitude of excrements by a diverse variety of vertebrates is constrained by the absence of any skeletal remains.

A later interpretation was very different, advancing the belief that all North American ferruginous “coprolite” localities represented intestinal casts of giant earthworms [53]. This hypothesis is yet another attempt to explain why a multitude of “bromalites” can occur in strata that contain no animal fossils. Weaknesses of the proposal include the assumption that an unknown type of giant aquatic earthworm existed from the late Mesozoic to the mid Cenozoic; although modern giant earthworms are live in Australia and South America [53], they inhabit terrestrial environments and are not aquatic. A second difficulty is that the meter-long cololites have coiled morphology, but the intestinal tracts of giant annelids are linear. Also, the great size range of the ferruginous specimens is a challenge. Broughton [53] speculated that 1–2 cm long specimens represented feces from the earthworms, and that multicm objects were defecations from unknown vertebrates. The combined hypotheses leave important issues unresolved, namely the absence of remains of vertebrates, and that absence of sediment bioturbation that would substantiate the presence of a multitude of giant annelids.

6.7. Diagnostic Criteria for Recognizing Coprolites

In summarizing the various hypothesis that have been advanced to explain the origin of ferruginous “bromalites”, the six diagnostic criteria listed by Amstutz [12] (1958) for recognizing coprolites provide a useful perspective.

• Extrusive forms: “There does not seem to be any process in geology which can produce extrusion shapes of this size or form” (p. 506, [12]).
• A flat base on the side of deposition.
• Striations typical of the great gut of mammals.
• Constantly limited length or quantity, typical for excrements.
• Variations in shape corresponding to variation in viscosity, as known from excrements
• Some forms correspond to the form of the great gut of mammals and were probably extruded as hard excrements.

These criteria appear to have been developed to support the preconceived hypothesis that Salmon Creek specimens are coprolites, a criticism first made in print by Spencer [58]. Our observations from of a multitude of Salmon Creek specimens reveals flaws in these conclusions.

• The Amstutz [12] statement is not based on observational data; it is merely a preconceived opinion. Abiogenic extrusion forms have been widely reported. One example is soft sediment deformation caused by biogenic methane production [59].
• Most Salmon Creek specimens do not have a flattened base; they are 3-dimensional. (Figure 1, Figure 6, and Figure 7) 3-dimensional forms are also a common characteristic of ferruginous “bromalites” from Madagascar and Saskatchewan (Figure 2 and Figure 3).
• Striations can be produced in any soft material that is extruded through an aperture that has rough edges (e.g., a hollow plant stem or a knot hole). It is not a characteristic unique to the great gut of mammals.
• Constantly limited length or quantity (mass) is not a characteristic of Salmon Creek ferruginous specimens. The Amstutz [12] observations were based on specimens that were subject to biased collecting.
• Viscosity is not a diagnostic characteristic for fecal excrement. For a single individual, viscosity may vary in accordance with state of health or diet, e.g., the excremental differences between constipation and diarrhea. Also, digestive processes are highly variable among different animals. Among herbivores the large, moist excrements associated with ruminant digestion (e.g., bovid) are very different from the small, firm pellets produced by elk, deer, and rabbits. Although extrusion characteristics of soft material are strongly affected by viscosity and plasticity, there are no consistent diagnostic characteristics for fecal masses.
• This statement fails to consider the rheology of extrusion: A muscle-controlled sphincter is not required to produce extruded masses that have pointed ends. When a plastic material is extruded through a narrow opening, the terminations may be controlled by surface tension. The effects of surface tension are evidenced by the convex shape of a drop of fluid on a smooth surface. For extrusions, the object shape is likely to be determined by a pressure gradient. This can be demonstrated by soft materials excreted from aerosol cans or squeeze tubes (Figure 43). In the case of aerosol cans (e.g., shaving cream), depressing the tip button results in the extrusion of soft material that initially emerges at relatively low pressure, which quickly increases to produce uniform flow. When the spray button is released, pressure does not instantly cease. Instead, a rapid pressure decrease results in a tapered shape caused by the surface tension of the extruded material. A similar phenomenon occurs with extrusions produced with squeeze tubes (e.g., toothpaste, cake frosting). The person squeezing the tube is likely to begin and end the extrusion process with gentle pressure, pointed ends again resulting from the effects of surface tension.

Injection (extrusion) of plastic material into wet, plastic sediment could produce tapering terminations via an additional process. This could apply to deformation of material as a result of the accumulation of methane, following a buried plant stem or similar pathway. The initial pressure needs to be high enough to create a pathway for the emerging gas, producing a conduit whose diameter would undergo dilation as the pressure increased. Conversely, as gas pressure decreased at the conclusion of the injection episode, the diameter of the conduit in the plastic sediment would be expected to decrease. The result is the formation of a mass with tapered terminations, independent of surface tension effects.

6.8. Comparative Studies

Comparative actualistic study is a tool that has long been absent from investigations at sites where ferruginous “bromalites” occur in enormous quantities. For example, Salmon Creek specimens have been extensively marketed as turtle coprolites, despite their dissimilarity to excrements from extant Testudinaea [60,61] (Figure 44).

6.9. Hypotheses for Abiogenic Origin

The first detailed geologic studies of the Wilkes Formation strata at Salmon Creek, Washington, were made by Roberts in 1958 [35]. He acknowledged the possibility that some of the extruded shapes were coprolites, but suggested that they were more likely to have resulted from plastic deformation of iron-rich sediment that had been squeezed through the roots and stems of aquatic plants. Danner [62,63,64] described the specimens as having originated from nonbiogenic extrusion of plastic masses of sedimentary siderite. The association with sideritic masses and non-mineralized fossil wood resulted in the hypothesis that the extrusion of soft sediment occurred when moist volcanic ash was pushed through fissures in submerged hollow logs [14,65]. This interpretation was amplified in a later report [66].

Mustoe [15] suggested that ferruginous extrusions at Salmon Creek were perhaps the result of soft sediment deformation triggered by the release of methane produced by microbial degradation of buried plant matter. Evidence of abiogenic origin was later presented by Yancey et al. [37,38]. These speculative interpretations contravene the common preconception of “I know poop when I see it”.

7. Summary: The Bromalite Enigma

The origin of ferruginous “coprolites” and “cololites” remains unprovable, and the variety of conflicting hypotheses are largely based on conjecture. There are several important elements that remain unresolved.

• Despite the occurrence of immense numbers of extrusive ferruginous specimens at localities that range in age from Paleozoic to Quaternary, with the exception of the Poland locality [17] there have been no discoveries where typical coprolite features are present. These characteristics include the presence of inclusions of undigested material, and the occurrence of other fossils in the strata where the ferruginous objects occur.
• The sizes and shapes of the ferruginous specimens have great diversity. The apparent uniformity of specimens is the result of collecting bias. At Salmon Creek, the majority of in situ specimens are very small.
• The common association of ferruginous extrusions with unmineralized wood requires explanation.
• The presumption that organic-rich excrements have been completely replaced by ferruginous minerals requires an explanation that allows wood in the same stratum to remain intact.
• The possibility that the ferruginous extrusions resulted from abiogenic processes is uncertain, because of the lack of discovery of any environment where objects of similar morphology and composition are being created. Precipitation of siderite and various iron oxides and hydroxides is common in the sedimentary record, and pressures created by methane or other gases are well documented. However, there are no known situations where these processes have combined to produce bromalite-like morphologies. In summary, there are no established models for either biogenic or abiogenic production of ferruginous “bromalites”.

Previous interpretations have commonly been based on limited data and authors emphasizing data that support a predetermined hypothesis. The reliance on limited numbers of specimens collected by other people and the absence of field work has been a recurring phenomenon. The identification of Salmon Creek ferruginous specimens as coprolites began in 1958 with Amstutz [12], who examined about 50 specimens that were collected near the Highway 99 Jackson Highway Bridge comprising material that had been fluvially transported from upstream strata. Since then, supporters of the coprolite interpretation have relied on small numbers of Salmon Creek specimens (e.g., [5,25]). In contrast, every field-based investigation has resulted in the conclusion that the extruded objects are probably pseudofossils [15,35,38,62,63,64].

Much of the evidence in this report supports the possibility of abiogenic origin for Salmon Creek “coprolites”, but a single depositional model cannot be applied to all occurrences. The published records are rife with contradictory assessments. In the face of these uncertainties, it is not possible at the present time to make definitive interpretations for the origin of ferruginous extrusions, which may be coprolites, cololites, or pseudofossils. This is not a search for the cure for a rare disease, or an effort to rescue astronauts trapped in space. Interpreting the origin of fossil excrement is a compelling topic for ichnologists, but it is not a mystery that demands a quick solution.

Despite their various imperfections, previous investigations have value as stepping stones leading to future reference. For example, researchers have presented analytical methods and petrologic examinations that can be applied in the future to other localities. I suggest that for the interpretation of the origin of ferruginous “bromalites”, this is a time to continue to follow the path of stepping stones rather than to declare arrival at a final destination.