Introduction

Elemental and isotopic signature analyses are routine in the biological and environmental sciences, being used as indicators of diet [1,2], ecotoxicology [3], pollution [4], soil sciences [5] and animal movement studies [6,7]. Current analytical techniques include stable isotope analysis (SIA), inductively coupled plasma mass spectrometry (ICP-S), atomic absorption spectroscopy (AAS), and neutron activation (NAA) among others [8]. However, many of these techniques are destructive, time consuming, and expensive, precluding their use in the field, or on live organisms. Development of rapid, in situ, inexpensive and non-destructive sampling techniques would greatly expand the utility of elemental signatures in biological and environmental sciences.

X-ray fluorescence (XRF) is a well-established methodology for measuring the elemental composition of samples [9]. It is a quantitative, non-destructive technique that measures the abundance of an element based on the characteristic emission of secondary x-rays following excitation by a primary x-ray beam [10]. Portable x-ray fluorescence (pXRF) instruments provide opportunities for field-based data collection. Developed initially for use in the field of geology [11], they are also used in a diverse range of fields including archaeology [12–14], art history [15,16], forensics [17,18] and food sciences [19,20] among others [21–23].

Data obtained from XRF is also used for a range of applied biological questions including the detection of in vivo disease for assessment in humans [24,25], toxicology in humans [26–28] and wildlife [29,30], and geographic provenance for wildlife [7,31]. However, it’s use in animal studies, has been limited relative to the opportunities it offers.

The portability, non-destructive and rapid multi-variable data collection offers significant opportunities in the field of animal sciences. However, due to its initial development for use in geology there are assumptions about sample characteristics that are not necessarily met by biological, and other non-geological samples. This has potential implications for the way in which samples should be processed and scanned, how raw data is processed by pXRF on-board algorithms, and the resulting data provided to the user.

These assumptions include the thickness and density of the sample, referred to as ‘infinite thickness’, i.e., the sample is of sufficient thickness and density that all x-rays are contained within the sample and do not pass through it [32,33]. It has been recognized that data acquired from samples that are thin, and do not meet infinite thickness need to be corrected [33]. The thickness required to meet this assumption varies with the density of the material being analysed [34], which is challenging to accurately determine in living samples.

Moisture content is another variable that can impact on the results returned by pXRF [35,36]. Water absorbs the characteristic x-rays from the elements and causes the primary radiation from the excitation sources to scatter. This results in a decrease in the intensity of characteristic x-rays and an increase in the intensity of scattered x-rays in a fluorescence spectrum [37]. Higher moisture content in soils has been shown to impact data resulting in lower elemental concentrations [35]. Moisture content of biological samples is highly varied and dependant on sample state, and in living samples moisture content is very difficult to determine.

The length of time a sample is exposed to x-rays can impact on the data obtained. Scanning times may affect the ability to detect low-concentration or low atomic weight elements, where longer scan times improve the signal-to-noise ratio [38]. However, in a biological context, there may be a trade-off between detecting elements, and exposing a live organism to excessive radiation. As such, determining whether elemental signatures significantly change in biological samples as a function of scanning time is a key consideration to optimise data quality and organism welfare.

Further, the location of the scan on the sample can also influence results. The partitioning of elements in biological samples may occur due to tissue type, metabolic processes [39], age, and sex [40]. Studies by Nganvongpanit et al. (2016) found different elemental concentrations for different bones within the same individual while Buddhachat et al., (2016) found differences between tissue types with keratinous tissues (hair, nail, skin), being different to bone.

The use of portable X-ray fluorescence (pXRF) as a non-destructive technique for elemental analysis of biological samples offers significant opportunities in the field of biological and environmental sciences. However, when planning a sampling design for biological samples, there are a number of potential sources of variation that need to be accounted for to ensure data are comparable and research questions are answerable. This paper aimed to explore the application of pXRF for elemental analysis of biological samples and to test the major sources of variation across four representative taxa, namely infinite thickness, sample state (dried/thawed), scanning time, and scanning location on the sample. The findings of this study provide insights for future studies and will contribute to the development of standardised protocols for pXRF analysis of biological samples.

Materials and methods

pXRF data collection

An Olympus Vanta M-Series portable x-ray fluorescence (pXRF) instrument (4-watt X-ray tube with tungsten (W) anode, 8-50keV, silicon drift detector), with three beam energies (10, 40 and 50keV) was used to scan all specimens. Where size permitted, specimens were scanned within the Olympus workstation [41]. Samples were scanned using the GeoChem3 method which uses a fundamental parameters algorithm [42] that automatically corrects for inter-element effects [41]. The instrument was factory calibrated by Olympus and calibration checks were performed throughout the sampling process using the onboard calibration option. The Vanta pXRF provides two output data types; raw beam spectra containing information across 2048 keV bands, and elemental concentrations as a percentage for 42 elements calculated via Olympus’ on-board algorithms.

Reptile, mammal, and fish specimens were scanned in both thawed and dried states, feathers were scanned in their natural state. All specimens were scanned at 20, 40, 60, and 80s/beam with and without silica backing at multiple locations (Fig 1) and states (Fig 2) with 2 replicate scans. The silica dioxide block backing consisted of 30 mm diameter x 25 mm solid cylinder. Prior to scanning, specimens were defrosted at room temperature for ~12 hours and scanned thawed. Specimens were then oven dried (60°C for 72 hours) and re-scanned.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Scanning locations (black dots) on each specimen of representative taxa.

https://doi.org/10.1371/journal.pone.0297830.g001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Summary of the number of individuals specimens per variation test and condition; natural state, thawed or dried.

https://doi.org/10.1371/journal.pone.0297830.g002

Results

We excluded 22 elements from our analyses (Y, Nb, Ag, In, Sb, Te, La, Ta, W, Pt, Au, Pb, Bi, Si, V, Cr, Co, Ce, Ne, Pd, U and Th) due to absences in their Kɑ1 and/or Kß2 peaks.

Infinite thickness assumptions were met for thawed and dried states for fish, rabbits and lizards using concentration data (Fig 3). Infinite thickness was not met for feathers using either concentration and beam spectra data (Fig 3).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Summary of results for each test using concentration data.

https://doi.org/10.1371/journal.pone.0297830.g003

Overall, we detected a mix of significant (p<0.05) and non-significant (p>0.05) differences in concentration data due to sample state (Fig 3 and S1 Table), scan time (Fig 3 and S2 Table) and scan location for all taxa (Fig 3 and S3 Table).

All three taxa (reptiles, fish, and mammals) tested for differences between dried and thawed sample states found significant differences in the concentration data overall, with multiple per-element differences identified (Fig 3 and S1 Table).

Scanning time had a significant effect on overall detected elemental concentrations for all taxa when comparing the 20 second beam time to the 40, 60 and 80 second beam times. (Fig 3). All three taxa tested for differences between scan locations revealed some significant differences in overall elemental concentration data amongst differing scan locations (Fig 3).

Overall device precision was very high, with a Spearman’s correlation coefficient of 0.999 when comparing replicate scans). There was variation in the correlation of replicate scans across elements, with heavier elements less correlated than lighter elements (S4 Table).

Discussion

In this study we investigated the application of pXRF for sampling biological samples, focusing on the effects of sample thickness, sample state, scanning time, and scan location on elemental concentration data. Our results found that sample state, and scan location resulted in the most significant differences between element concentrations. Infinite thickness assumptions were met for most tests regardless of sample state, excluding feather samples. Scan time results found in most cases the 40, 60 and 80 second beam times were equivalent.

These findings are consistent with previous studies that reported the impacts of sample moisture [35,50], and sample thickness [34,35] on pXRF data collected from soil samples. Extended scan times were not found to result in significant data gains, which is supported by findings by Williams, Taylor [51], however is contrary to findings by Zhang, Specht [38], which found extended (5 minute) scan times reduced variability in results when scanning bone through soft tissue. Extending scan times on live specimens increases radiation exposure which can result in health impacts [52]. Estevam and Appoloni [24] calculated that a 50 second scan using an x-ray source of 13 and 17 keV resulted in an exposure of 3 mSV.

Based on the results of this study and considering animal care and ethics we would recommend when scanning live specimens that scan times are kept to a minimum while still ensuring adequate data is obtained, our study suggests that 40 s/beam is sufficient to achieve this for most elements (Fig 3). We would also recommend scanning a part of the body that is of sufficient thickness to meet infinite thickness assumptions, but away from key organs to limit the impact of radiation exposure, and to scan with a standardised and consistent backing behind the specimen whenever possible. For the Olympus Vanta this is the provided silicon dioxide block but any other inert, elementally simple material should suffice. This study found that on thawed specimens (akin to live organisms with regards to body thickness and moisture) the majority of scan locations met infinite thickness requirements (Fig 3). However, noting that different scan locations on the specimen may results in different results (Fig 3). This is an important consideration if wanting standardised data between specimens. To ensure consistency the same location should always be scanned. Lastly, noting that we found significant differences in data due the state (thawed or dried) of the sample.

There are some limitations of this study that should be considered when translating our results. Firstly, we only tested a limited number of biological samples from a small set of taxa. Our results may not be generalizable to other types of biological samples but are intended as a guide to the impact different sample characteristics, and sampling techniques may have on results. Secondly, we used a single pXRF instrument with a specific configuration and settings. Different instruments or settings may produce different results [53], however the broad effects demonstrated here should hold as the sources of variation are extrinsic to the instrument. Finally, we did not evaluate the long-term stability or repeatability of pXRF measurements for biological samples [12]. Future research should address these issues by testing more diverse biological samples and elements, comparing different pXRF instruments and settings, and assessing the quality control and assurance procedures for pXRF analysis.

This study demonstrates that pXRF can be a useful tool for biological research if sample characteristics and sample design are cognisant of the assumptions of the pXRF device, and differences within and between samples as highlighted in this study. The advantages of pXRF over conventional analytical techniques include its portability, non-destructiveness, speed, and capability to measure many elements simultaneously. These features make pXRF an attractive option for biological studies that require in situ or large-scale analysis of elemental signatures in various types of samples, such as feathers, hair, bones, plants, soils, etc. However, our study also highlights the considerations required when sampling biological samples using pXRF and the potential impact on results obtained, which include where to scan, how long to scan and how to prepare the sample.

Supporting information

Acknowledgments

B. Jackson, Murdoch University, WA. P. Meagher, Taronga Conservation Society, D. Harasti, NSW Department of Planning and Environment, The Reptile Lab, Macquarie University.