Introduction

The trace fossil Rosselia Dahmer1 consists of a spindle shaped structure with a mostly vertical central shaft surrounded by concentric, fine-grained laminae composed of material collected on the sea floor by a detritus-feeding organism2,3,4 (Fig. 1). Rosselia is common in stressed shallow-marine environments such as delta fronts or storm-influenced shoreface deposits throughout most of the Phanerozoic e.g.,5,6,7,8,9. Because the laminae consist of fine-grained, organic-rich material, they also act as recorders of the organic sedimentary component in sandy environments where the preservation potential for organic matter is otherwise limited. Here we present high-resolution sediment geochemical parameters across two Rosselia specimens collected from the early Pliocene Kueichulin Formation (Fig. 2) in western Taiwan (Fig. 3). We use these data to test the hypothesis that the concentric layers act as short-term, high-resolution environmental recorders. We then assess how diagenetic alteration of the trace fossil influences the geochemical signals in the concentric layers. Finally, we discuss the value of Rosselia as paleo-environmental recorders through the Phanerozoic.

Figure 1
figure 1

Field photo of Rosselia and conceptual illustration of its formation. (A) Cartoon illustrating how a marine bottom-feeder constructs a Rosselia burrow by ingesting detrital material at the sea floor and then deposits the resulting fecal material around the central shaft. In settings with rapid sediment deposition, this often leads to an upwards adjustment of the spindle (equilibrichnion) (adapted from2). (B) Photo of a vertical cross-section (top to the left, bottom to the right) of the lower part of Rosselia, illustrating the sand-filled shaft and distinct mud laminae. The top of the trace fossil is cross-cut by crustacean burrow (yellow arrow).

Full size image
Figure 2
figure 2

Stratigraphic chart for the studied region after22. Red star marks the stratigraphic position of the Rosselia samples analyzed in this study.

Full size image
Figure 3
figure 3

Geological maps of the study area based on31,32 (map drawn using https://www.adobe.com/adobe/illustrator). (A) Map showing the general geology of Taiwan. Black box indicates the region of panel B. (B) Geological map of the study region. Red star indicates sampling location of Rosselia analyzed in this study.

Full size image

Rosselia occurs most commonly with trace-fossil assemblages typical of the Skolithos and Cruziana Ichnofacies. More recently, Rosselia became the namesake for the Rosselia Ichnofacies, which is considered to be typical of sandy, delta-front environments9. While the largest Rosselia may reach lengths up to one meter2 and diameters of more than 15 cm, typical sizes are 10–20 cm in length, and a 2–10 cm in diameter1,2,10. The concentric layers surrounding the central shaft comprise material that is considerably finer than the surrounding host substrate, and the central shaft is commonly filled with sediment similar in composition to the host substrate. Alternating colors of concentric laminations are the result of differences in grain size or sediment composition2,11.

Rosselia form through the repeated accretion of fine-grained material on the sides of the vertical shaft. As new material is added, older layers are pushed into the surrounding sediment resulting in the deposition of concentric laminae surrounding the shaft. The oldest laminae are furthest away from the center, and increasingly younger laminae occur towards the center of the spindle-shaped trace fossil (Fig. 1). The probable trace maker for Rosselia is a terebellid polychaete2,3,4, and the laminae in the burrow are believed to have formed as a result of the terebellid’s bottom-feeding habit: their tentacles were utilized to accrete sediment from the sea floor onto the inner wall of the burrow during the process of consumption. Modern terebellid worms often gather their food on the sea floor from the opening of their dwelling tube, resulting in a thickened tube wall at depth12. However, some polychaetes are known to convey large amounts of organic material collected on the sea floor into the sediment without ingestion13.

The fact that laminations in Rosselia are added sequentially over time suggests that they could act as high-resolution environmental recorders registering, for example, geochemical changes in the composition of fine-grained seafloor material collected by the trace-making organism. The concentric laminations of Rosselia would thus be similar to tree-ring records, but with the oldest layers furthest from the center of the burrow. The Rosselia tracemaker has been shown to survive severe erosional events caused by larger storms14, and individual animals lived for several years. Indeed, modern terebellid polychaetes have life spans that range from months to several years15,16. As well, Rosselia is a common trace fossil with a long stratigraphic range from Cambrian17 to the present7,18. Taken together, analysis of concentric laminations of Rosselia could provide snapshots as well as short, but highly resolved time series of variations in environmental conditions throughout the Phanerozoic. For example, in a paleogeographic setting such as the early Pliocene of proto-Taiwan, the steep topography of the rapidly uplifting orogen19,20,21 and heavy monsoon rains likely resulted in distinct seasonal alternations in the input of terrestrial vs. marine sourced organic carbon22,23. These seasonal swings would then be recorded as distinct shifts in the δ13C values of organic matter that make up the concentric layers.

Diagenetic processes may also influence and change the geochemical and isotopic signals recorded in the laminae. For example, isotopic studies on carbonates in the concentric layers of Rosselia revealed symmetrical variations in δ18O, indicating that the isotopic composition had been altered by a series of cementation events11.

Results

Morphological description

Several specimens were collected from the Kueichulin Formation where it is exposed along the Da’an River in western Taiwan (see Figs. 2 and 3). Two Rosselia specimens are analyzed (Da’an-3 and Da’an-4), and both show an elliptical outline in horizontal cross-section, with the longest axis being about 70 mm in both cases. In both specimens, the concentric laminae are made up of alternating dark gray, greyish, and light grey layers that display a symmetrical distribution on both sides of the central shaft. Observed lamina thicknesses typically range from 0.5–1 mm, although individual lamina up to 6 mm were observed. These may, however, be the result of an amalgamation of several laminae. Thin-section analysis shows that while the central tube and surrounding host sediment are dominated by closely packed quartz grains, the dark colored concentric laminae are dominated by a dark matrix with only sparse quartz grains. In contrast, light colored concentric laminae contain high numbers of quartz grains, and greyish layers contain an intermediate number of grains (Fig. 4). This suggests that the color of the concentric laminae is controlled by the ratio of fine-grained matrix to coarse-grained particles. In Da’an-4, diagenetic alteration of the outmost ~ 1 cm of the specimen has resulted in a distinctly colored rim surrounding the entire trace fossil. Detailed thin-section analysis of this diagenetic rim show that its composition does not differ significantly from the inner, unaffected concentric layers.

Figure 4
figure 4

Polarized microscope images of mineral distributions and structures in thin sections derived from the Daan-4 specimen. Quartz is the dominant mineral of silt- and sand-sized material, with rare feldspar. The mineral grains are mostly angular and poorly sorted, and cemented by dark-colored matrix. (A) Cross-section view of the Daan-4 specimen. (B) Boundary between the central shaft and lamina, showing a high abundance of quartz grains in the central shaft (lower right corner), light-colored lamina (band of quartz grains stretching from left middle to upper right), and dark matrix-filled areas reflecting dark-colored laminae (upper left corner and a matrix-filled band between central burrow and light-colored lamina). (C) Boundary between unaltered (upper right corner) and diagenetically altered (lower left corner) laminae; the color of the mud matrices differ due to diagenetic alteration. (D) Diagenetically altered lamina showing the same pattern as unaltered laminae: high abundance of quartz grains reflect light-colored laminae, while matrix-filled bands reflect dark-colored laminae. (E) Central shaft and surrounding laminae; the central shaft is dominated by quartz grains. (F) Host rock (sandstone); dominated by densely packed quartz grains. (G) Boundary between diagenetically altered laminae and host rock.

Full size image

XRF core scanning

Elemental variations along transects through the two Rosselia specimens show a conspicuous correspondence between most elements and the concentric laminae. In particular, high concentrations of Si, typically related to mineral grains, correspond to the light- and greyish- laminae. Elevated concentrations in Ti, Rb, and K occur in the dark laminae, suggesting that clay minerals are an important component in these laminae (Fig. 5a, b). Multivariate analysis of elemental variations through the two trace fossils confirms this qualitative comparison. The elements group into distinct clusters corresponding to the central tube, the different concentric layers, and the surrounding host sediment. From the multivariate analysis, it is also clear that the light concentric laminae and central tube (blue cluster) are dominated by Si, and that darker laminae (yellow and pink clusters) are dominated by elements present in clay minerals (Ti, Rb, K). Moreover, the yellow clusters are specifically rich in S and Ca (Fig. 5c, d).

Figure 5
figure 5

X-ray fluorescence data (normalized) of the Da’an-3 and Da’an-4 specimens derived using an Itrax core scanner. Principal component analysis (PCA) biplots of the XRF data of Da’an-3 and Da’an-4. Only elements with high reproducibility were selected as variables for PCA and cluster analysis (determined through duplicate scans), resulting in a total of 12 selected variables: Si, Mn, Fe, K, Rb, Ti, Cr, Zr, Zn, S, Ca, Y. Results of cluster analysis are presented in the figures through colors. In both Rosselia samples, the blue cluster overlaps with lighter laminae (including central burrow) and show high amounts of Si. Both the pink and yellow clusters overlap with darker laminae, showing high amounts of elements present in clay minerals (Ti, Rb, K).The yellow cluster is also rich in S and Ca. In Da’an-4, it is clear that the yellow cluster is darker in color than the pink cluster, and its green and beige clusters match the outer diagenetically altered laminae. The number of clusters was determined through the calculation of silhouette width, while cluster significance was verified through the multiple response permutation procedure (MRPP) and analysis of similarities (ANOSIM).

Full size image

Both visual inspection of the elemental variations along the transect and the multivariate clusters show a generally symmetric distribution of darker and lighter laminae on both sides of the central shaft, although a number of discrepancies are evident. First, the outer, older laminae are more often discontinuous or difficult to trace around the trace fossil. Second, redox sensitive elements, such as Mn and Fe, display a conspicuous pattern in the outmost diagenetically influenced rim of the Da’an-4 specimen. Peaks at the edges of the diagenetically influenced laminae suggest that Mn, and to some extent Fe, have been mobilized from inside of the laminae and migrated towards the boundaries where it reprecipitated. This movement of cations resulted in strongly elevated Mn, and elevated Fe. For comparison, the Da’an-3 specimen lacks visual indications of diagenesis, and both Mn and Fe concentrations exhibit similar trends in concentrations to the other minerogenic elements.

Total organic carbon and δ13C

Total Organic Carbon (TOC) in both specimens shows considerably higher concentrations in darker laminae compared to lighter laminae, and both the host rock and central shaft contain even lower TOC. TOC content in dark laminae typically ranges from 0.5 to 0.9% (average 0.72%), and in lighter laminae TOC ranges from 0.4 to 0.75% (average 0.61%). Diagenetically overprinted laminae in Da’an-4 have TOC content that range from 0.3 to 0.7% (average 0.56%), although one outlier has a TOC > 1% (Fig. 6). Overall, the TOC contents of the diagenetically altered burrow margin in Da’an-4 is 0.1–0.2% lower than the light-colored and dark-colored unaffected laminae.

Figure 6
figure 6

Boxplots of total organic carbon (TOC) values and stable carbon isotope ratios (δ13C) in dark-colored unaltered laminae, light-colored unaltered laminae, all unaltered laminae, and all diagenetically altered laminae of sample Da’an-4. Blue diamonds show values of all measured data points.

Full size image

Similarly, there is a general tendency for lower δ13C values to occur in light-colored laminae (average − 26.4‰) compared to darker laminae (average − 26.2‰). Interestingly, the δ13C values of the diagenetically altered laminae in Da’an-4 (average − 25.7‰) are approximately half a ‰ higher than the δ13C values of the unaffected laminae (Fig. 6).

Statistical evaluation (Kruskal–Wallis Test) showed a statistically significant difference between dark and light laminae, as well as between unaltered and diagenetically influenced laminae for both TOC and δ13C (Supplementary data 1).

Discussion

The general symmetry of the concentric laminae of the two studied Rosselia specimens observed both in thin sections and in elemental distribution measured using XRF core scanning suggest that each lamina represents one feeding season for the producer. Based on fair-weather deposits from an inner-shelf setting in the Western Gondwana Paraná Basin, southern Brazil, Netto et al.5 suggested that Rosselia is a stress-tolerant burrowing adaptation to environments often affected by erosional events in which the trace-making organism collects fine-grained material mobilized during storms (e.g., hyperpycnal flows). This interpretation is corroborated by the close association of Rosselia and sandy delta-front environments9. In fact, equilibrium response to erosional and depositional events by the Rosselia producers are known at least since the Ordovician24. Nara25 postulated that the enhanced influx of organic material during erosional events would provide beneficial conditions for the Rosselia trace-maker.

Consequently, Rosselia in the Kueichulin Formation in western Taiwan should preferentially record geochemical signals of material settling out after tropical cyclones. Unfortunately, despite the general symmetry of the laminae on both sides of the central shaft, it is rarely possible to systematically trace all laminae around the entire trace fossil. The paleoenvironmental signal recorded in the burrows, therefore, represents a series of snapshots rather than a continuous sequence.

Carbon isotope composition of the organic material in Rosselia are in the same range as those reported for the Kueichulin Formation by Dashtgard et al.23 who concluded that the organic material deposited in the early Pliocene proto-Taiwan Strait was almost entirely terrestrially sourced, and delivered primarily during tropical cyclones. The small shift (0.2‰) in δ13C values observed in the dark, fine-grained laminae (average − 26.2‰) vs light coarse-grained laminae (average − 26.4‰) could indicate a slight preference for the ingestion of marine-sourced organic-matter deposited in dark laminae. Presumably, the trace fossil producer had a preference for labile marine sourced carbon collected on the sea floor during quiet times over the more refractory carbon delivered by hyperpycnal flows from rivers cf.26,27. Additional support for an increased marine influence in the darker laminae comes from the distribution of the elements S and Ca. Both elements are largely of marine, biogenic origin cf.28, and show enhanced levels in the darker laminae, but considerably lower levels in lighter laminae and in surrounding host rock. Nevertheless, despite the small differences between light and dark laminae, the range of the isotopic composition recorded in Rosselia laminae suggest these structures provide information on shifting organic carbon sources in settings where the sandstones themselves contain very low TOC.

A limiting factor influencing the usefulness of δ13C trends derived from Rosselia is the diagenetic overprint. In specimen Da’an-4, the outmost approximately 1 cm of the trace fossil displays a diagenetically altered rim. Carbon isotope values from this rim consistently displayed approximately 1 ‰ higher δ13C values compared to values from inner, unaffected laminae. Consequently, care must be taken to not sample material for organic carbon isotope analysis that shows visible evidence of diagenetic alteration.

From the discussion above, the following conclusions regarding the use of Rosselia as a paleoenvironmental recorder are drawn:

  • The concentric laminae in Rosselia are deposited in a semi-continuous way, providing a snapshot of environmental conditions on the sea floor.

  • The dark, fine-grained laminae of Rosselia can be a source of information on paleo-sea floor values of δ13C in coarse-grained settings where TOC is otherwise too low for accurate measurements.

  • Diagenetic processes may significantly alter both geochemical and isotopic signals recorded in the concentric laminae.

Because variations in the composition of sedimentary organic matter delivered to deltas and nearshore shallow marine environments reflect changes in the river catchment, organic matter can provide insights into processes such as provenance shifts due to orogenic processes, or paleoenvironmental changes caused by climatic shifts. Unfortunately, preservation of organic matter is usually poor in sandy environments. Here the preference of Rosselia for stressed, sandy environments cf.9, and the incorporation of substantial amounts of fine-grained, organic rich material provide a powerful tool to assess long term changes in the type and volume of organic matter delivered by ancient river systems. The long temporal range of Rosselia spanning most of the Phanerozoic9,10 allows this tool to be applied to a wide variety of scientific problems.

Methods

Two samples of Rosselia preserved in fine-grained sandstone (Da’an-3) and coarse-grained sandstone (Da’an-4) host rocks, were taken from the early Pliocene Yutengping Sandstone Member of the Kueichulin Formation and analyzed in this study. Each sample was cut into several slabs perpendicular to the central shaft, exposing the concentric lamination of the spindle shaped part of the trace fossil. Each slab was then used for mineralogical and geochemical analyses.

Three analyses were applied to each sample: thin section analysis, X-ray Fluorescence (XRF) core scanning, and organic carbon analysis.

Thin sections of about 30 µm (μm) in thickness were made using Buehler’s PetroThin instrument, and focused on the central tube, the concentric laminae, and the host rock directly surrounding each Rosselia sample. Structural and mineralogical characteristics were then determined through observation of the thin sections in a polarized microscope.

The Itrax Core Scanner conducts XRF elemental analysis and produces semi-quantitative results in elemental content29,30. A slab from each Rosselia sample was scanned using the core scanner along straight lines that passed through the central burrow, perpendicular to the laminae, and into the host rock on either side. The Itrax has an analytical footprint of 0.2*4 mm and the samples were scanned with a step size of 0.2 mm. X-ray tube parameters include exposure time (5 s), voltage (30 kV) and current (50 mA). Duplicate scans were done to compare signal stability for each of the elements. The resulting elemental data was normalized before further statistical analyses were executed. Elements that proved unreliable due to the instrument design (Al, Ar, Mo, Ce, and W) and non-matching signal patterns between the duplicates were removed from the analysis. Observation points with uncommonly low overall signal strength and/or high Ar signal values most likely indicate cracks in the sample, were also removed. Cluster analysis and principal component analysis (PCA) were performed using the cleaned data to determine relative elemental fluctuation inside the Rosselia samples. The number of clusters was determined through calculating silhouette widths of each cluster and choosing the combination of highest values as the optimal outcome, while cluster significance was verified through the multiple response permutation procedure (MRPP) and analysis of similarities (ANOSIM).

The fluctuation of organic carbon in the laminae of each Rosselia sample was analyzed. Powdered samples were collected from each lamina by hand using a dentist drill, and samples were then soaked in 1 N HCl to remove carbonates. Total organic carbon (TOC) content was measured by an element analyzer (vario MICRO CUBE, elementar), and carbon isotope composition represented as δ13C value was analyzed by using an elemental analyzer (Flash EA, Thermo) connected with an isotope ratio mass spectrometry (Delta V, Thermo Finnigan).