The relationship between the phosphate and structural carbonate fractionation of fallow deer bioapatite in tooth enamel

Rationale The species‐specific relationship between phosphate (δ18OP values) and structural carbonate (δ18OC values) oxygen isotope ratios has been established for several modern and fossil animal species but until now it has not been investigated in European fallow deer (Dama dama dama). This study describes the relationship between phosphate and structural carbonate bioapatite in tooth enamel of extant fallow deer, which will help us further understand the species' unique environmental and cultural history. Methods The oxygen isotope composition of phosphate (δ18OP value) and structural carbonate (δ18OC value) of hydroxylapatite was determined in 51 modern fallow deer tooth enamel samples from across Europe and West Asia. The δ18OC values were measured on a GV IsoPrime dual‐inlet mass spectrometer and the δ18OP values on a temperature‐controlled elemental analyser (TC/EA) coupled to a DeltaPlus XL isotope ratio mass spectrometer via a ConFlo III interface. Results This study establishes a direct and linear relationship between the δ18OC and δ18OP values from fallow deer tooth enamel (δ18OC = +9.244(±0.216) + 0.958 * δ18OP (±0.013)). Despite the successful regression, the variation in δ18O values from samples collected in the same geographical area is greater than expected, although the results cluster in broad climatic groupings when Koppen‐Geiger classifications are taken into account for the individuals' locations. Conclusions This is the first comprehensive study of the relationship between ionic forms of oxygen (phosphate oxygen and structural carbonate) in fallow deer dental enamel. The new equation will allow direct comparison with other herbivore data. Variable δ18O values within populations of fallow deer broadly reflect the ecological zones they are found in which may explain this pattern of results in other euryphagic species.


| INTRODUCTION
A recent programme of inter-disciplinary research has shown that the distribution of fallow deer (Dama dama dama) is a direct record of human migration, trade, behaviour and worldview. [1][2][3][4] Prior to this understanding of the cultural significance of fallow deer, the species has been under-investigated in favour of the red (Cervus elaphus) and roe (Capreolus capreolus) deer, i.e. those that are native to western Europe. Yet the very presence of fallow deer in areas beyond their native eastern Mediterranean habitat, transported and maintained by human groups, demonstrates that investigations of their habits, habitats and provenance are important to palaeoenvironmental, human-animal and archaeological studies.
Strontium isotope analysis has been used to great effect to infer ancient translocations and source populations of fallow deer; 2,3 however, these issues could be addressed with greater resolution if they were complemented by oxygen isotope analysis. As first proposed by Longinelli,5 an important application of oxygen isotope biochemistry is paleoclimate reconstruction from fossil bone and tooth enamel. This technique has been used to examine place of origin for humans and animals in the archaeological record, as the isotope signature of local water sources is preserved at the point of mineralisation in dental enamel. [6][7][8][9] Within mammalian tissues including teeth, antler, ivory, and bone, bioapatite [generalised as Ca 10 (PO 4 ,CO 3 ) 6 (OH,CO 3 ) 2 ] contains two ionic forms of oxygen suitable for isotope analysis: structural carbonate (CO 3 2− ) and the more abundant phosphate (PO 4 3− ) (hereafter referred to, respectively, as O C and O P ). [10][11][12][13][14] Most published studies of bioapatite oxygen measure the δ 18 O C value as it is quicker, easier and cheaper to measure than the δ 18 O P value, 6,15 and so the relationship between the δ 18 O P and δ 18 O C values has been established for a range of modern and fossil animal species. [15][16][17][18][19][20][21][22] These data suggest that the relationship between δ 18 O P and δ 18 O C values (slope, intercept and Δ C-P ) is species specific, 21 where Δ C-P is the difference between the δ 18 O C and δ 18 O P values. Until now this relationship has not been investigated in the widely distributed and culturally important fallow deer.
The isotope ratios of oxygen atoms in O C and O P are cogenetic, as they are formed simultaneously in isotopic equilibrium with body water oxygen. This is directly related to the composition of ingested water, often meteoric water, at a constant body temperature, 5,18,[23][24][25][26] which is sensitive to latitude, altitude and climate. 27 Thus, if the δ 18 O values of tissues are measured, the results can help to assess an animal's origin and movement. 20,28,29 The relationship between the δ 18 O P value and the δ 18 O value of drinking water (hereafter the δ 18 O DW value) is also known to be species-specific and therefore it can also be part of provenance studies. 10,17,18,20,30 While this is well established in humans 5,23,31,32 and a range of animals including several deer species (examples in 10,33 ), it is currently unknown for fallow deer.
A sample of eight fallow deer from a single area of Italy was included in a larger study of red deer δ 18 Table 1). As a result, our samples broadly represent the geographic, climate and topographic range of human-introduced fallow deer distribution across Europe and West Asia. 35 The distribution of fallow deer has been significantly influenced by humans to the extent that its distribution now spans six continents. [1][2][3][4]35 In general, when human-mitigated circumstances allow, they live in a variety of deciduous and mixed woodlands, but rarely thrive at heights of more than 1000 meters above sea level. Fallow deer spend much of their time grazing in nearby fields, keeping close to wooded areas for cover and shelter but also for browse. 35 As the samples were drawn from herds managed in different ways this study reflects the range of complex and varied relationships that fallow deer have with people in their culturally defined environments. Table 1 and Figure 1 summarise information on the location and circumstances of each of the deer herds.

| Intra-individual variation
The teeth used in the study were permanent dentition from adult animals. Wherever possible 3 rd mandibular molars were collected; however, for three locations (France, Haifa and Turkey) only 1 st mandibular incisors were available (

| Sample preparation
The enamel preparation method for all analyses (cutting and mechanical cleaning) is after Montgomery. 44 A section of crown surface was abraded to a depth of >100 μm using a tungsten carbide dental bur and the removed material discarded. A thin slice of enamel was then cut from the tooth using a flexible diamond-edged rotary dental saw.
While sequential sub-sampling is frequently used to investigate seasonal climatic fluctuations and/or movement of herbivores, 42,[45][46][47] bulk enamel samples were used in this study. These larger sections represent an "average" isotope ratio, which relates indirectly to the average meteoric water δ 18 O value during tooth formation, i.e. a period of several months to several years. 42 As the year/month of birth, age and year/month/season of death data for many of the wild animals included in this study are unknown, these values provide an average signature for each tooth during formation. To provide consistency across the samples, the same section of tooth was measured in each individual. The most worn 3 rd molar from across the sample was measured at 6 mm from the cervix of the crown to tip. Each 3 rd molar sample from other individuals was cut to reflect this. None of the 1 st incisors showed signs of wear in the same way and so the full length of the tooth was sampled in each case.
All sawn surfaces were mechanically cleaned with a tungsten carbide dental bur, and any adhering dentine was removed. The enamel chips were cleaned ultrasonically for 5 min in high-purity water and rinsed twice to remove loosely adhered material. This method ensured that any surficial contaminants were removed.    see Table 3 Sample   48 The 1σ reproducibility of the KCM reference material for this set of analyses was calculated by analysis of variance (ANOVA), that separates the within-batch variation from the between-batch variation. 49 The results of the ANOVA for within-batch repeatability for δ 18 O C and δ 13 C C values were ±0.08‰ and ±0.04‰, respectively.

| Isotope analysis of oxygen in structural carbonate (δ 18 O C value)
The between-batch reproducibility was statistically insignificant compared with the within-batch repeatability. with an isotope ratio of 2.0052‰. 52 The within-batch repeatability for B2207 by ANOVA produced a p-value of 0.23, while the betweenbatch reproducibility was statistically insignificant by comparison.

| RESULTS AND DISCUSSION
The results of the analyses presented in Table 2

| The regression
The linear relationship between phosphate oxygen and structural carbonate oxygen in fallow deer enamel was determined by regressing the data using a Functional Relationship Estimation by Maximum Likelihood (FREML). 55  where the p-value is 0, r 2 = 0.8736, the 95% confidence interval is 1.54σ for n = 51 and the values within brackets are the standard error ( Figure 4). where the p-value is <0.001, r 2 = 0.956, the 95% confidence interval is 1.39σ for n = 43 and the values within brackets are the standard error ( Figure 5).
The second regression calculation omitting the eight samples that lay below the regression line has a better correlation coefficient.
However, to avoid overlooking potentially important natural variability in fallow deer populations, the rest of the investigation refers to the first regression, determined with all 51 fallow deer specimens included.  Figure 6A). These data, which cover similarly broad geographic ranges to the fallow deer in this study, have comparable slopes (between 0.963 and 1.044) and variable intercepts (between −7.7‰ and −9.7‰).

| Fallow deer and other regressions
Further comparisons show that regression data for mixed fossil species from Germany produced a lower slope of 0.892 and an intercept of −3.788 (r 2 = 0.66, n = 49). 20 Archaeological human samples from the UK 6 have correlation equations that also fall within the range of Tütken et al. 20 No specific studies have previously been carried out on deer species; however, we extracted deer data from larger studies: modern red deer data from Iacumin et al, 18 Italian fossil red deer data from Pellegrini et al, 57 and German extinct Miocene deer from Tütken et al. 20 We used these to calculate the relationship between δ 18

| Inter-tooth variation
In the case of four individuals, we were able to examine the differences between 1 st incisors and 3 rd molar teeth (Table 2 and Figure 7). None of these individuals were deemed outliers in Δ 18 O by our statistical tests.
The data show no consistent difference between teeth for δ 18 O P , δ 18 O C and δ 13 C C values. This is based on a small sample set; however, the differences between teeth in the same individual are no more or less significant than those seen between each of the three individuals from the 13 geographic locations in this study. This has significant implications for sampling strategy, and while we continue to consider the results from 1 st incisors and 3 rd molar teeth together in this investigation, we acknowledge that this requires further testing to see if this has further bearing on our results and interpretations.

| Fallow deer δ 18 O variation
Despite the successful regression confirming the relationship between  Table 2).
These variations are greater than was expected for animals with the same physiology collected in the same area (see error bars, Figure 8).
It was expected that the tooth samples from the same sites,  (Tables 2 and 3). 60 We found that fallow deer δ 18 O values cluster well within these climate groupings ( Figures 9A and 9B). This means the δ 18 O composition of fallow deer, and other non-obligate drinking mammals, may be seen as a broader environmental indicator than that of obligate drinking species, which may be more specific in pinpointing provenance based on drinking water. It is important to consider why this may be the case.
In the wild, fallow deer home ranges are thought to be ca 9.75 km 2 in males and 2.1 km 2 in females 61 although Chapman and Chapman 35 have suggested that this can be wider during mating excursions. Within the managed environments represented by the deer in this study, these ranges can be altered or restricted according to different management strategies: fallow deer in a zoo environment will have very different ranges from those in national parks or hunting reserves. Within the more 'natural' ranges, such as larger parks and hunting reserves, landscape and relative humidity may vary, particularly with changes in elevation, but it is also likely that plant availability may vary. Some of the individuals in the study, including those in the zoo, are likely to be more intensively managed by regular or supplemental feeding, but whether or the extent to which this happens may depend on climatic conditions in a given season, or may vary in terms of which foods are used.
While received wisdom suggests that fallow deer are primarily grazers, it has been shown that they will feed on many things including browse, fungi and fruits, as they are opportunistic in their foraging habits. 62 Table 3     and white-tailed deer. 36,58 As fallow deer have variable δ 18