A case study in forensic soil examination from China

Soil can often be found and collected in different kinds of cases, and provides an important form of trace evidence in forensic investigations (e.g. Murray & Tedrow 1992; Pye 2007; Ruffell & McKinley 2008; Murray 2011; Fitzpatrick 2013a, b). Because of its widespread nature and ease of transfer, in many circumstances it can provide information to link people to crime scenes. The evidential value of soil depends on the variation in its characteristics. In China, soil has many varieties, which differ in colour, texture, minerals and organic matter. Diversity of soil strongly depends on natural and anthropogenic factors, such as topography, climate, botanical and microbiological functions, conditions of watering, and human activities. As well, there are a large number of soil features that are invisible to the naked eye, such as pollen, spores, and artificial or synthetic materials (e.g. fibres). One of the key tasks in forensic soil investigations is to compare questioned sample items such as digging tools, footwear or clothing with samples taken from known locations of interest to the enquiry, such as a crime scene (Fitzpatrick & Raven 2016). By measuring the physical, chemical, mineralogical and biological properties the of the soil, together with several other related natural and anthropogenic substances, critical forensic information can be obtained about the location and human activities of a suspect (e.g. Collins et al. 1997; Ingram & Lin 2002; Rawlins et al. 2003). The forensic examination of soil is not only concerned with the analysis of naturally occurring rocks and minerals (e.g. Saferstein 1998), it also includes the examination of vegetation, organic matter, pollen grains and spores, which sometimes will impart the soil with characteristics that make it unique to a particular location and its surrounding environment. Many papers have been published reporting on soil examinations that involve specific methods such as elemental analysis (e.g. Reidy et al. 2013; Woods et al. 2014), mineral identification (e.g. Hirokawa et al. 2016), soil botany (e.g. Ferri et al. 2015) and pollen examination (e.g. Morgan et al. 2014; Preusche & Weber 2014). However, few papers have reported on using combined multiple techniques to examine soil samples in a crime investigation. Multiple examinations can provide much more information in different dimensions for soil source tracing and comparison.

In the countryside of, Jilin Province, in the NE of China, a dead body was found lying near a stream. Soil was found adhering to the clothing. Based on the soil morphology characteristics, such as colour and particle size, it was apparently different from the soil located at the stream site. It was concluded that the body was likely to have been transferred from another place. After examination, the soil sampled from the body was suspected to have originated from a ginseng planting field, which helped the police to find the burial site. It was 15 km away, to the east of the stream where the body was found. The soil samples from the burial site and some reference soils were collected to make comparisons to assist in finding and prosecuting the suspect. Soil evidence in this investigation was regarded as crucial evidence in convicting the suspect of murder if even without the suspect's DNA being available.

About 0.1 g of soil was collected from the body and was observed under a microscope. Even with the naked eye, plenty of tiny plant debris could be identified in the soil sample. The species of plant debris was likely to provide critical information about the surrounding environment of the place where the soil came from. This information would also be vital in locating the initial burial site. Plant deoxyribonucleic acid (DNA) barcoding is a relatively new technique that uses DNA sequences from a small fragment of the genome to identify a plant species. This DNA technique has not been widely used to identify plant species in soils for forensic purposes in China. In this homicide investigation, the DNA barcoding technique was carried out to determine the species of the plant debris in the soil sample.

Two pieces of plant debris were handpicked from the soil, and were labelled as 1A and 2A. The debris was then crushed to powder using a mortar and pestle. Genomic DNA was extracted from the powdered sample using a DNeasy Plant Mini Kit (Qiagen) following the manufacturer's protocols. High-quality DNA analyses results for samples 1A and 2A were obtained. Polymerase chain reaction (PCR) amplification was performed for two regions: ITS2 and rbcL. The primers for the ITS2 and rbcL were tested and the general PCR reaction conditions were obtained from previous studies (Chen et al. 2010). Purified PCR products were sequenced in both directions with the primers used for PCR amplification on a 3730XL sequencer (Applied Biosystems, USA).

To estimate the quality of the generated sequence traces, the original forward and reverse sequences were assembled using CodonCode Aligner 3.0 (CodonCode Co., USA). The nucleotide sequence alignment based on the BLAST search and plant sequences was downloaded from GenBank. The Kimura 2-Parameter (K2P) distances of the candidate sequences were computed using MEGA V5.1 software to create a neighbour-joining (NJ) tree.

The NJ tree obtained for plant debris 2A is shown in Figure 1. It shows that the ITS2 and rbcL are all clustered together with Panax ginseng. The same result was obtained for sample 1A, which clearly indicates that the plant debris in the soil sample originated from Panax ginseng. This strongly indicated that the soil could come from a place where ginseng was planted. As a result of the efforts made by the police, the first burial site of the body was located at the suspect's ginseng plantation.

The control soil sample was collected from the burial site by the police. Several reference soil samples were also collected to confirm the results of the soil examination. Ginseng plants are usually planted in a series of lanes, as shown in Figure 2a. Besides the sample collected from the burial site in lane 1 by the police, five more soil samples were collected near the burial site along lane 1 (denoted by the yellow stars in Fig. 2a). In addition, five samples were collected along lane 2 next to lane 1 (denoted by the red stars in Fig. 2a). Four other reference samples in the ginseng plantation were also collected individually 10 and 20 km apart. A total of 14 reference soil samples for measurement were obtained, which are detailed in Table 1. The relative position of the burial site, the ginseng plantation fields 10 km away and the ginseng planting fields 20 km away are also shown in Figure 2b.

Mineral identification, and elemental and pollen analyses were carried out to make more definitive comparisons between the questioned and control soil samples. Three types of analyses were conducted on all of the reference soil samples.

All soil samples were ground and sieved. X-ray diffraction (XRD) analysis was carried out to identify the mineralogy in all of the soil samples. The sieved fraction with a particle size of <74 µm was used to prepare samples for the XRD analysis. The mineralogy examination was carried out on a Multiflex model XRD instrument (Rigaku). X-ray powder diffraction data were collected in the angular range 5°–90° with a step 0.02° and a 5 s counting time. The XRD patterns of the questioned and control samples are shown in Figure 3. The results indicated that quartz and feldspar were the main mineral phases present in all samples. All soil samples presented the following very similar results: quartz, vermiculite, chlorite, hydromica, tremolite and feldspar. Although it was true that the mineralogy results did not assist in narrowing down the origin of the questioned soil sample, these results did correspond to the expected minerals found in the soils in this region.

The same sieve fraction with a particle size of <74 µm was used to make pressed disks for XRF examination. XRF measurements were carried out on a ZSX100e (Rigaku) wavelength-dispersive XRF spectrometer with an X-ray tube with a 4 kW rhodium (Rh) target. The area analysed on the soil samples was a spot 3 mm in diameter and the measuring atmosphere was in vacuum. The elements detected and their relative weight percentages in all soil samples are listed in Table 2. Fourteen kinds of elements were detected in all soil samples. It was apparent that the elemental compositions for Na, Al, Si, P, S, Ca, Fe and Sr in samples R3-1and R3-2 (shown in bold type in Table 2) was different to those of the questioned samples and the control samples. Hierarchical cluster analysis was used to assess the homogeneity between the questioned sample and the other samples based on relative weight percentages. The results showed that questioned and control samples, CS1 and CS2, collected near the burial site and R2 were clustered together, while R3 was totally different to the others. This demonstrated that elemental analysis was useful in differentiating between the soil samples. The hierarchical cluster result is presented in Figure 4.

Pollen analysts have long used a standardized methodology to extract pollen from sediments and soils based on the density difference between pollen spore and soil in a heavy liquid (e.g. Moore et al. 1991). The heavy liquid used was a mixture of potassium iodide, hydriodic acid and arsenic-free zinc particles, which was adjusted by mixing with water until the density value was about 1.9 g cm⁻³. The pollen floated in the heavy liquid and was decanted off by centrifuging, whereas the heavier inorganic soil particles (such as quartz and clay) sank to the bottom. As a consequence, the pollen from the soil samples was concentrated more efficiently. This has proven effective in concentrating the pollen from soils, and permits a comparison of pollen assemblages from different regions and various environments.

Slides were then made and examined using a light microscope. At least 400 spores and pollen grains were counted for each sample under the Leica DM 2500 light microscope at a magnification of ×400.

The main components of pollen assemblages of algae, fungi, Sinopteridaceae, Athyriaceae, Abies, Picea, Pinus, Betula, Gramineae and Tilia were found in the questioned and control soil samples. Photographs of the different kinds of pollen in the questioned soil sample under the microscope are shown in Figure 5. The pollen assemblages corresponded exactly to the surrounding area, complying with the characteristics of pollen expected to be found in the woods of the NE part of China.

Pollen types identified and their number percentages (counting at least 400 pollen grains per slide) in all soil samples are shown in Figure 6. The questioned and control soil samples had different pollen assemblages compared to the 10 reference soil samples (i.e. R2-1, R2-2, R3-1 and R3-2 samples). This indicated that the four soil samples collected from the surrounding environment were different to the soils collected at the burial site. Reference samples CS1 and CS2 demonstrated almost the same pollen assemblages as the questioned and control soil samples. The result showed that soil samples collected from a small area presented similar pollen assemblages.

Taking into account all of the soil forensic examination results, it was confirmed that the questioned soil sample was identical to the control soil samples and the reference soil samples (CS1 and CS2) collected from surrounding areas, but different to the other four reference samples (R2 and R3), which were located further away from the burial site. Multiple techniques (plant fragments, pollen, mineralogy and elemental analysis) helped to determine the origin of the recovered soil and to make comparisons between soil samples. Elemental and pollen analysis showed higher resolution.

The authors are grateful to Professor Wang Guiqiang and Wang Yufei for many valuable discussions. We also thank the reviewers and editor for assistance.

The case study was supported by The National Key R&D Program of China (grant 2017YFC0803803 to H. Guo).

HG: Methodology (Lead), Project administration (Lead); PW: Data curation (Equal), Formal analysis (Equal); CH: Data curation (Equal), Formal analysis (Equal); JZ: Data curation (Supporting), Formal analysis (Supporting); XY: Data curation (Equal); YQ: Data curation (Equal), Formal analysis (Equal); HM: Data curation (Supporting); JL: Data curation (Equal).