Mammoth teeth have been widely investigated using stable-isotopic analysis
for paleoenvironmental and paleoecological reconstructions due to their
large size and frequent discoveries. Many past investigations sampled the
tooth enamel with the “bulk” method, which involves drilling one sample from
the occlusal surface to the root for each tooth. Some of the more recent
studies applied the “sequential” method, with a sequence of samples drilled
following the dominant enamel growth direction to produce a time series of
isotopic oscillations that reflects high-resolution environmental changes, as
well as changes in mammoth dietary behavior. Although both the bulk and
mean sequential
Aufgrund ihrer relativen Fundhäufigkeit und Grösse werden Mammuthmolare vielfach für paläoökologische Zwecke und Umweltrekonstruktionen mit Hilfe von stabilen Isotopen herangezogen. Bei vielen der bislang publizierten Arbeiten wurde dabei Zahnschmelz mit der “Bulk”-Methode beprobt, bei der für jeden Zahn eine einzelne Probe von der Kaufläche bis zur Wurzel gebohrt wird. Neuere Studien wenden nun die “sequenzielle” Methode an, bei der eine Reihe von Proben entlang der Hauptwachstumsrichtung des Zahnschmelzes gebohrt wird, um die Variation der Isotopenwerte über die Zeit zu bestimmen, welche sowohl hochauflösende Umweltveränderungen als auch Veränderungen im Ernährungsverhalten der Mammuts widerspiegeln kann. Obwohl sowohl “Bulk”- als auch die gemittelten sequenziellen
Stable-isotopic analysis on mineralized tissues of animals has added great knowledge to our understanding of past environments and climates. Among all the animal tissues, Pleistocene mammoth teeth and tusks are of special interest for paleoenvironmental and paleoclimatic reconstructions over the past decades due to their large size and frequent discoveries. Oxygen-isotope ratios in mammoths can represent their surrounding environmental properties because they are directly related to the isotopic ratios of their ingested environmental water, which in turn primarily reflects regional temperature and water balance (Dansgaard, 1964; Longinelli, 1984; Luz et al., 1984). Previous studies of oxygen isotopes in mammoth remains have provided paleoenvironmental records in Europe and North America from Marine Isotope Stage (MIS) 5 to MIS 2 (130–22 ka cal BP). In most previous studies, the sampling method has been drilling one “bulk” sample from each tooth from the occlusal surface to the root (Genoni et al., 1998; Tütken et al., 2007; Ukkonen et al., 2007; Iacumin et al., 2010; Kovács et al., 2012; Pryor et al., 2013). The purpose of bulk sampling is to try to cover the longest possible period of time of tooth formation (Pryor et al., 2013), as the sample should represent an averaged isotopic signal across the tooth formation time (Fricke and O'Neil, 1996; Sharp and Cerling, 1998; Hoppe, 2004). Although this method can effectively reconstruct the averaged paleoclimatic conditions over several years, the temporal resolution of reconstruction is limited to decadal scale, and consequently, a very small amount of data exist on sub-annual environmental conditions and climatic variations during the Quaternary from these regions. Such data, however, are crucial in understanding how the highly variable climate of the Late Pleistocene translated into regional- to local-scale environmental conditions, ultimately affecting a range of animal–environmental and human–environmental interactions (Denton et al., 2005; Bradtmöller et al., 2012; Prendergast and Schöne, 2017; Prendergast et al., 2018).
To address this issue, a “sequential” approach has also been applied to mammoth and isotope research in some studies (Koch et al., 1989; Fisher and Fox, 2007; Metcalfe and Longstaffe, 2012; Widga et al., 2021; Wooller et al., 2021). Like most mammal species, mammoth tooth enamel has a dominant growth direction from the occlusal surface to the roots at a relatively constant rate (Metcalfe and Longstaffe, 2012). Therefore, it has the potential to yield highly resolved time series of paleoenvironmental information over the course of tooth formation. In several recent studies, multiple sequential samples were drilled from the same mammoth tooth following its growth direction with a resolution up to 1mm per sample, forming a time series of isotopic oscillations that likely reflects paleoenvironmental changes at sub-annual scales (Metcalfe and Longstaffe, 2012; Wooller et al., 2021; Miller et al., 2022). The mean value of the sequential samples should therefore reflect the averaged isotopic signal during the period of time of tooth growth, which is also what the bulk sample is expected to represent. However, this remains untested, and it is unknown whether the mean sequential value and the bulk value obtained from the same mammoth tooth yield similar isotopic compositions, as well as whether they have recorded the environmental properties during approximately the same period of time. Due to this limitation, it is still uncertain whether isotopic results obtained from bulk and sequential sampling methods can be directly compared, interpreted, and used for paleoenvironmental reconstruction.
In this study, we explored the potential differences between the sequential
and the traditional bulk sampling methods. We obtained four mammoth teeth of
MIS 3 age from southwestern Germany and applied both sampling methods on
each tooth. The bulk and average sequential
Similar to modern elephants, the woolly mammoths (
A photograph and its corresponding schematic diagram of a mammoth molar tooth used in this study (UW-1).
Within each molar plate, mammoth tooth enamel has primary and secondary
stages of formation (Smith, 1998; Smith and Tafforeau, 2008). During the secretory (primary) stage, daily incremental features grow from the enamel-dentine junction (EDJ) to the outermost surface, and these incremental lines are parallel to the EDJ under microscopic view (Metcalfe and Longstaffe, 2012). Tooth increments formed during the primary stage take
up only about 20 %–30 % of the entire enamel weight (Passey and Cerling, 2002; Passey et al., 2005). The maturation (secondary) stage starts after the secretory stage ends, and it takes up most enamel weight and formation time (approximately two-thirds of total formation time) (Smith, 1998). During this stage, enamel formation starts at the apex (occlusal surface) near the EDJ side, growing dominantly along the tooth height to the root while extending to the outermost surface simultaneously. Secondary daily incremental features captured by microscopic analysis indicated that enamel growth direction is inclined at an angle of 55–60
Oxygen isotopic composition in animal bioapatite can be used as an indicator
of past climates and environmental conditions. For large-sized homeothermic
animals (animals which can keep a constant body temperature) such as the
mammoths, the
Three complete molar teeth (one M1 and two M3 molars) and one fragmentary
molar of woolly mammoths (
Information on the mammoth teeth used in this study. Radiocarbon
dating was undertaken at the Research School of Earth Science, Australian
National University, and calibrated with Calib
(
A 1 mm cylindrical diamond-coated drill bit attached to a low speed hand
drill and a state drill model were used to drill the enamel and collect its
powder. Two sampling methods were applied. The first method was bulk
drilling, which involves taking one sample from one enamel ridge by scraping
down the entire enamel length, and only a thin layer (
We then conducted chemical pretreatment, following the protocol developed by
Snoeck and Pellegrini (2015), which involves soaking the samples in acetic acid (1 M, buffered with sodium acetate, pH
The similarity of bulk and mean sequential values was first assessed by
their offset. If the offset is less than the analytical precision of the
mass spectrometer (0.12 ‰), the two values are
considered the same, and
In addition to offsets, we also used a “modified seasonal method” to analyze where the bulk value sits in relation to the range of values from the
sequential samples. This method is based on the season of mollusk collection
method from Prendergast et al. (2016), in which four seasons can be
differentiated from the sequential records by generating quartiles in the
range of
To address the two issues, this seasonal method was further modified, and we
decided to calculate a “relative percentile” of bulk values compared to
sequential data distributions with the following equation:
In total, 24 bulk and 1114 sequential samples were drilled out of 24 enamel
ridges. The results for each enamel ridge are summarized in Table 2 and
plotted in Fig. 5. In the sequential results, we removed the outliers with
boxplots. The data distributions of each tooth were box-plotted separately,
and values that are smaller than the 25th percentile minus 1.5 interquartile ranges and greater than the 75th percentile plus 1.5 interquartile ranges were considered as outliers and removed (
The list of results of the study, including the length, number of
identifiable years, and the bulk and sequential
NA: not available.
Data distribution of the four teeth and comparison between bulk and
sequential results. Each cross represents the
For both bulk and average sequential values, more positive
The cyclicity of all enamel ridges was assessed to identify features of
seasonal environmental changes. For UW-1, UW-2, and UW-3, the intra-tooth
sequential
Examples of
Significant differences (
The relative percentile values were also analyzed and shown in Table 2 and Fig. 7. The bulk values of most enamel ridges of teeth UW-1 (4 out of 6) and UW-2 (6 out of 10) are located between the 25th and 75th percentiles within the middle quartile with the remaining values falling below the 25th percentile including one value for UW-2 being more negative than the minimal sequential value. The bulk values of all enamel ridges of tooth UW-3 are below the 25th percentile, with the majority (5 out if 7) being more negative than the minimal sequential value. The relative percentile of UW-3b is 96.3 %, the only data point above the 75th percentile.
The plot of relative percentiles of bulk values from all the enamel ridges of the four teeth.
It is essential to evaluate whether physical and chemical transformations of
sample materials may lead to isotopic exchange and loss of primary isotopic
signal (Keenan, 2016; Goedert et al., 2016). The preservation condition of our mammoth teeth was assessed, and based on three pieces of evidence, at least part of their original materials seems to have been preserved. The first piece of evidence is the sequential
Although there is still a possibility of diagenetic alteration, all samples
were pretreated with acetic acid to remove any secondary carbonates.
Additionally, due to its low porosity and low organic contents, tooth enamel
is one of the most resistant materials to diagenetic alteration and is
capable of preserving primary isotopic signals for millions of years. Based
on all the above considerations, we assume our mammoth teeth have at least
partially preserved their original isotopic signals, and the acquired
Differences between bulk and sequential results are present in all four
mammoth teeth, with the bulk values being generally more negative than the mean
sequential values. There is only one exception, which is the enamel ridge of
UW-3b, with the bulk value being 1.35 ‰ more positive. This may be caused by the absence of a complete annual cycle, as only 37.5 mm of material was sampled. The incomplete hydrological year in the records of UW-3b could also cause the data distribution of sequential samples to taper. Older tooth samples also have generally greater offsets. Only one enamel ridge from UW-1 has a difference less than 0.12 ‰, which is within the precision of the mass spectrometer. Therefore, the bulk and
sequential values for this sample can be considered as equal. Three enamel
ridges from UW-2 show slightly greater offsets between 0.12 ‰ and
0.15 ‰, which may still be considered as similar. Most
other enamel ridges in teeth UW-1 and UW-2 (
One explanation for the difference is the geometry of incremental lines. Our
two sampling methods each predominantly cover one growth direction: a bulk
sample covering the entire enamel height with only a thin layer of enamel
collected, whereas each sequential sample includes a deeper sampling pit
covering the majority of the enamel thickness but with only a small
proportion of the enamel height. Neither of the two methods followed the
exact enamel growth direction: the incremental lines of mammoth secondary
enamel material are inclined at an angle of 55–60
A simplified model showing secondary incremental lines formed
during different seasons and the sampling spots.
This problem may be more significant in enamel ridges of shorter lengths. In
Fig. 9, we plotted the difference values against the incremental lengths of each enamel ridge. Teeth UW-2 and UW-3 have enamel ridges of different lengths, varying from 32 to 112 mm, and both of them show strong negative
relationships between enamel length and difference value. No correlation was
detected from the enamel ridges of UW-1. This could be due to their similar
lengths (136–160 mm). Inter-tooth comparison also indicates that enamel
ridges with longer lengths generally have smaller differences between bulk
and mean sequential values: greater difference values (
Plot of difference values against incremental lengths of all enamel ridges.
The geometry of incremental lines may have another consequence, which is reduced seasonal amplitude in sequential isotopic records (Passey et al., 2005). In Fig. 8, one sequential sample may contain increments formed during more than one season, especially those formed during seasonal transitions. In fact, oxygen and carbon isotopic cyclicity has been detected from the EDJ to the outer surface, indicating that there was tooth growth through the enamel thickness, although the growth rate was much slower compared to the growth along tooth height. Therefore, each sequential sample obtained in this study may hold an averaged isotopic signal of one or multiple seasons, and the isotopic variation in environmental water would have potentially greater amplitude.
Another possible explanation for the difference between bulk and sequential
results is that the two methods collect enamel formed during different
times. Reade et al. (2015) sampled the tooth enamel of Barbary sheep
(
A third explanation is the mixture of primary and secondary enamel increments in the samples. We primarily sampled the secondary incremental lines in the teeth. However, there was also a primary mineralization phase in mammoth tooth formation, although it only took up 20 %–30 % of the total enamel mass (Passey and Cerling, 2002; Passey et al., 2005). Primary mineralization forms vertical incremental lines under microscopic view, indicating the material grows from the EDJ to the outer surface (Metcalfe and Longstaffe, 2012). Therefore, the thin layer of enamel powder collected by the bulk sampling method may also contain primary mineralized material formed during a short period of time. This period of time could be during any season; consequently, if the collected primary increments were formed during extreme summer or winter months, the mixture of primary and secondary enamel material might skew the isotopic signal to one direction. This could be the cause of the bulk values of several enamel ridges falling outside the data distribution of sequential values, since primary increments were formed during a different time to the secondary increments. If the collected primary increments were formed during spring or autumn months with the isotopic signal in environmental water closer to the annual average, there is less impact from primary material, and the bulk and mean sequential values would be more similar. The sequential samples again are less influenced by the primary mineralization, as more enamel thickness is sampled, covering primary increments formed during a longer period of time and maybe including multiple seasons.
The bulk and mean sequential results obtained in this study show both similar and different values. Here we discuss how these differences may help us decide the optimal sampling methods for future research. The bulk samples, at least those obtained through the method applied in this study, do not always reflect an averaged isotopic signal at decadal scales. Instead, it is more likely a reflection of either extreme summer/winter environmental conditions or spring/autumn isotopic compositions which are close to the annual mean value. In comparison, the sequential method is less influenced by the primary increments and the geometry of secondary incremental lines. We do not suggest the sequential sampling method itself can more accurately reflect paleoenvironmental conditions. The main reason it does in this study is that with our drilling method, the combination of all sequential samples covers the entire enamel length and the majority of enamel thickness, which included increments formed during a longer period of time than the thin layer of the bulk sample. Reade et al. (2015) suggested sampling the entire tooth height, as well as the enamel thickness, to reduce isotopic damping caused by the enamel growth pattern. We also recommend that future studies which involve surface sampling on mammoth teeth drill all available enamel material in both directions, regardless of which sampling method will be applied.
However, this means we can only apply one method to each enamel ridge. We
suggest the sequential sampling method is more optimal because it
provides not only an averaged multi-annual-scale isotopic composition (mean
sequential value) but more importantly a time series of high-resolution
paleoenvironmental records. Such records are crucial for understanding the
highly variable climatic conditions of the Late Pleistocene and deciphering
human–environmental interactions (Denton et al., 2005;
Bradtmöller et al., 2012; Prendergast et al., 2018). In addition, the
sequential
Although we recommend using complete and large-sized mammoth molar teeth as the preferred study material, we are aware that such samples are not always available at certain sites. Mammoth teeth may have been weathered and transported by surrounding environments, and consequently they may break into fragments and lose enamel thickness through erosion. Due to this limitation, it is unclear which part of the enamel and how much enamel thickness and tooth height were sampled in many past stable-isotope studies on mammoth teeth, especially those which employed bulk sampling methods. Given that the bulk values in some of our samples fall outside the range of sequential values, we recommend only comparing results obtained from sequential sampling methods with other sequential sampling studies. If results obtained from two different methods are compared directly, the potential isotopic offset between bulk and sequential samples may be falsely concluded to reflect different environmental conditions.
The main limitation of this study is the different locations of the two sampling methods. Although the sampling spots are from the same trenches, bulk and sequential samples are powder from different parts of the enamel ridge. A possible improvement in further studies is to use only part of the powder in each sequential sample and combine all the remaining powder along one enamel ridge to form the bulk sample. In this case, enamel powder taken from the two methods would have covered roughly the same incremental length of enamel thickness and tooth height. Another possible improvement is applying the two methods on the enamel ridges at two sides of the same molar plate. However, this method assumes that enamel ridges on both sides grow simultaneously. Nonetheless, this study has provided insights into the degree to which we can compare past studies, as it is likely that different parts of enamel ridges were sampled in different studies, and some studies might have only sampled a thin layer of enamel material as the bulk samples.
Another limitation is that we considered primary enamel mineralization as a factor that caused differences in bulk and sequential results, but we are uncertain about the proportion of primary mineralization collected in each sample and its degree of impact. These questions require further investigations into mammoth dental morphology and separate isotopic analysis of primary and secondary incremental materials.
In addition, although the sequential samples provide high-resolution records of likely seasonal environmental changes, they are not perfect reflections of paleoenvironments. The geometry of secondary incremental lines potentially caused several different seasons recorded in one sequential sample, averaging the isotopic signal and in turn reducing the amplitude of isotopic oscillations. In future sequential sampling of mammoth teeth, we may section the teeth to expose the enamel thickness, identify incremental features under microscopic views so that we can perform spot-drilling on each incremental line to avoid seasonal averaging.
We applied both bulk and sequential sampling methods to mammoth enamel and compared the results. Both similar and different
We also effectively obtained high-resolution paleoenvironmental records of sub-seasonal resolution. These records hold the potential of reconstructing paleo-climate and paleo-hydrology for various aspects, such as seasonality and inter-annual differences, in future studies. However, we must be aware that the amplitude of isotopic signal obtained from this sequential sampling method might be reduced, since the drilling direction has an angle against the incremental lines so that enamel materials formed during more than one season could be mixed in each sample. This issue may be improved by combining the sequential sampling method with microscopic analysis on mammoth dental morphology to achieve spot-drilling on each incremental line in future research.
The data that support the findings of this study will be uploaded to the PANGAEA database and be open to other researchers in the near future.
The research idea was conceived by JHM and AP; sample acquisition was carried out by JHM; experiment planning was carried out by ZL, AP, and JHM; laboratory work was carried out by ZL; isotopic data processing was carried out by RD; data analysis and interpretation were carried out by ZL, AP, and JHM; ZL drafted the manuscript. All co-authors reviewed and made inputs to the manuscript.
At least one of the (co-)authors is a member of the editorial board of
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This article is part of the special issue “Quaternary research from and inspired by the first virtual DEUQUA conference”. It is a result of the vDEUQUA2021 online conference in September/October 2021.
We would like to thank Frank Kiesewetter from the Ubstadt-Weiher gravel quarry for collecting and providing the mammoth teeth for this research, and we thank Rachel Wood from Australian National University for undertaking the radiocarbon dating of the samples. We are also grateful for Niels de Winter and another anonymous reviewer for providing valuable feedback.
This paper was edited by Christian Zeeden and reviewed by Niels de Winter and one anonymous referee.