Morphological identification: Each collected soil specimen was mainly examined under a Nikon SMZ745T stereomicroscope and photo-documented with an adapted Canon DS126761 camera. Features of larger specimens were documented with an Olympus TG-4 camera. For the polyxenid millipedes, specimens were prepared, stained, mounted and observed under high magnifications of a compound microscope as previously described (Short and Huynh 2010). For further examination, whole polyxenid specimens were observed under scanning electron microscope (SEM) as previously described (Huynh and Veenstra 2017). In brief, samples were first preserved in 80% ethanol and dehydrated by passing them through a graded series of ethanol (80%, 90% and 100%). The samples were then bathed in acetone for 2 min and air-dried for a further 2 min. Each specimen was subsequently mounted on a stub for gold coating using a Fisons sputter coater (0.02 mbar, 18 mA, 2 nm min^–1), then examined using a JEOL (JSM – IT300 Scanning Electron Microscope). Digital SEM images of the specimens were obtained for documentation.

DNA extraction and barcoding: After photo-documentation of specimens, tissues from either the head or body were dissected and blotted on tissue paper to remove excess ethanol. Genomic DNA from these tissues was isolated by a spin-column based extraction method using the PureLink™ Genomic DNA Mini Kit (Invitrogen, USA), following the manufacturer's instructions. The remaining body parts of the specimens were transferred back to 95% ethanol and stored at -20ºC. The extracted DNA was subjected to quality and quantity control by 1% gel electrophoresis and One/Onec Microvolume UV-Vis Spectrophotometer (Thermo Scientific, NanoDrop, USA). The qualified genomic DNA was subjected to polymerase chain reaction (PCR) using a pair of universal primers (LCO1490: 5'-GGTCAACAAATCATAAAGATATTGG-3' & HCO2198: 5'-TAAACTTCAGGGTGACCAAAAAATCA-3') for amplifying mitochondrial cytochrome c oxidase subunit I (COI) gene of all collected samples (Folmer et al. 1994). Additional markers, including 18S ribosomal RNA (F: 5'-CTGGTTGATCCTGCCAGT-3' & R: 5'-TATTGATCCTTCCGCAGGTTCACCT-3') and 16S ribosomal RNA (16a: 5'-CGCCTGTTTATCAAAAACAT-3' & 16b: 5'-CCGGTCTGAACTCAGATCATGT-3'), were also used for barcoding millipede species (Folmer et al. 1994Giribet et al. 1996). PCR was carried out on a T100™ thermal cycler (Bio-Rad, USA) with the following parameters: an initial denaturation step at 95°C for 3 minutes; followed by 36 amplification cycles of 30 seconds for denaturation at 95°C, 30 seconds for primer annealing at 43-52°C and 35 seconds for extension at 72°C and a final extension step at 72°C for 5 minutes. The reaction mixture included PCR buffer, extracted genomic DNA sample, 2 mM dNTP, 1.5 mM MgCl₂, 0.4 mM of each forward and reverse primers and Taq DNA polymerase. The amplified products were then validated by 1% agarose gel electrophoresis and sent to BGI Genomics Company Hong Kong for Sanger sequencing on the platform ABI3730xl. The Chain Termination PCR (CTPCR) technique was used for sequencing the targeted amplified DNA fragment. In brief, the sent unpurified PCR products were first isolated by magnetic beads and the purified products were used as templates for CTPCR. CTPCR resembles the standard PCR, but with the addition of fluorescent labelled dideoxyribonucleotides (ddNTPs), instead of the conventional nucleotides. The resultant oligonucleotide copies synthesised were subjected to gel electrophoresis for size separation. The gel was then passed through the computer sensor to read the fluorescent signals emitted from the terminal ddNTPs of each oligonucleotide copy, thus producing a chromatogram and generating a full nucleotide sequence of the input DNA.

Molecular identification and sequence analysis: The chromatogram of each barcode was examined base by base on the software SnapGene Viewer. Manual deletion of primer sequences at the 5' and 3' ends was performed for each barcoded sequence. The resultant sequence was submitted to NCBI GenBank for homology sequence search. If there were existing barcodes that matched at least 99% of the input sequence, the species identity was assigned to the specimen. If the match score was below the threshold (99%), it meant that we have produced the first sequence for the species that was not previously barcoded. All the barcodes were then submitted to NCBI and the accession numbers provided are listed in Suppl. material 2.

Soil Macrofauna Composition: A total of 3,588 individual samples were collected from October 2019 to October 2020, including 150 different morphospecies. (Fig. 2Suppl. materials 3, 4). The sampled individuals belong to three major phyla (Arthropoda, Annelida and Mollusca) and eight classes (Arachnida, Chilopoda, Diplopoda, Symphyla, Oligochaeta, Gastropoda, Insecta and Malacostraca). The sampled soil communities were mainly dominated by Diplopoda and Oligochaeta which accounted for 23 and 15 out of 150 collected morphopspecies, respectively (Figs 2, 3).

Figure 2.  

The 150 soil macrofauna morphospecies collected in this study. The labelled numbers indicate the number of morphospecies in each animal group collected in this study.

Figure 3.  

The most abundant soil organisms from each macrofaunal order and their distributions (The New Territories (NT) and Kowloon (KL); Hong Kong Island (HK); and the islands (Is)). The blackened squares indicate the presence of animals in corresponding collection sites on the map.

Figure 4.  

Biodiversity of millipedes revealed in this project. A The composition of different millipede orders found in Hong Kong; B A Venn diagram showing millipedes found in the three major study areas; C The distribution of millipede orders in the three major study areas.

Amongst the 1,440 collected millipede samples, 23 millipede morphospecies were identified (Fig. 5), comprising 8 out of 16 extant known millipede orders (Fig. 2). The millipede community is dominated by Polydesmida and Spirobolida, which accounted for 10 and 5 out of 23 collected morphospecies, respectively (Fig. 4Suppl. material 3). The greatest number of millipede species can be found in New Territories and Kowloon (23), followed by HK Island (14) and other islands (12). Ten millipede species could be commonly found amongst these three major areas, while eight millipedes could only be found in the New Territories and Kowloon (23). It is also worth noting that one millipede species, the sphaerotheriid Zephronia profuga, could only be found on the Hong Kong Island (Fig. 4). In addition, Monographis queenslandica (originally found in Australia) and Alloproctoides remyi (originally found in Reunion and Mauritius) are two polyxenid species that have never been reported to be present in Hong Kong and this study first demonstrated their existence in Hong Kong.

Figure 5.  

Twenty three millipede morphospecies documented in this project.

Soil analysis: The pH value and electrical conductivity of each soil slurry sample (soil:distilled water = 1:2.5 w/v) were measured by a pH meter (EA940, Orion Research Inc., USA) and conductivity meter (ION6+, Oakton Instruments, USA), respectively. After extraction of soil samples using 1 M of ammonium acetate (pH 7), the concentration of exchangeable cations (Na, K, Mg, Ca) were then determined by either atomic absorption spectroscopy (SpectrAA-200, Varian, USA) or inductively coupled plasma optical emission spectroscopy (5800 ICP-OES, Agilent, USA). For the organic matter, soil samples were gravimetrically combusted at 550ºC for 4 hours according to the loss-on-ignition method, whereas the total Kjeldahl nitrogen and total phosphorus content were determined by semi-micro Kjeldahl digestion and acid digestion, while the latter was then measured by UV-visible spectroscopy (UV-1800, Shimadzu, Japan).

Soil properties in winter and summer: when averaged across all the sampling sites, the mean soil pH was 5.64 ± 0.08 (1 standard error) in winter and 5.82 ± 0.08 in summer, which was acidic and typical of the soil acidity in the local environment. The mean soil electrical conductivity was 297.77 ± 20.30 µS/cm in winter and 316.92 ± 25.88 µS/cm in summer, which was indicative of a general salt-free environment. The mean soil organic matter content was 6.99% ± 0.26% in winter and 9.27% ± 0.78% in summer, which was in the moderate range and considered suitable for plant growth. The average percentages of clay, silt and sand particles were 15.0%, 16.0% and 69.0%, respectively in winter and 14.53%, 12.28% and 73.19%, respectively in summer, leading to a sandy loam texture. The mean total Kjeldahl nitrogen concentration was 0.24% ± 0.01% in winter and 0.34% ± 0.04% in summer, which fell within the medium range for plant growth. The total phosphorus concentration in soils was 754 ± 70.35 mg/kg in winter and 1545.85 ± 361.30 mg/kg in summer, which was higher than the values of 300-500 mg/kg reported in some natural subtropical Chinese forests. This also implied a great seasonal variation in total phosphorus concentration (Suppl. material 5).

Before the beginning of this project, the understanding of soil biodiversity in Hong Kong, including the understanding of its contained millipede species, was inadequate. Previous studies have identified native millipede species, including Trigoniulus corallinus Gervais, 1847, Anaulaciulus tonginus Karsch, 1881, Hyleoglomeris bicolor Wood, 1865, Zephronia profuga Attems, 1936, Glyphiulus granulatus Gervais, 1847, Glyphiulus formosus Pocock, 1985 and Helicorthomorpha holstii Pocock 1895, to be present in local soil ecosystems (Shelley and Lehtinen 1999Mikhaljova et al. 2011Evenhuis and Eldredge 2010Golovatch et al. 2011Pocock 1895Golovatch et al. 2007Wesener and Koenig 2016). Yet, the surveys and information were given quite a time ago and there is no current comprehensive update on local biodiversity. Therefore, a citizen-approach survey was conducted in this study. Throughout a year of survey on 21 sites of urban and semi-natural habitats, this project has identified 23 millipede morphospecies alongside 127 other soil macrofauna in Hong Kong. For some of the chosen samples, a total of 646 DNA barcodes, including COI, 18S, rRNA and 16S rRNA genes, were performed and assigned to the specimens (from different regions), which significantly increases our understanding of the soil biodiversity in this region. We are aware that the reference barcoding sequences of myriapods on GenBank are far understudied compared with other arthropods. The barcodes generated in this study have provided additional and first sequences of millipedes that were identified in Hong Kong. This study demonstrates the need for morphological identification and molecular barcoding in contributing to the research of understudied animal groups.

Amongst the collected soil morphospecies, millipedes and earthworms are dominant. After reviewing the collected samples, we agree that there might be some collection bias throughout the survey process. Since the collection were mainly done by citizens through hand-picking methods, there is a tendency for the non-experts to collect those soil animals that are relatively large, conspicuous and slow-moving. Thus, this somehow explains a slightly biased collection effort within particular groups, neglecting some of the other soil macrofauna, including spiders and centipedes. In addition, the time (60 minutes) for on-site soil surveying that was adopted in the current study might not have provided enough time for isolation and collection of soil organisms. In the future, a mass of soil body should be collected and adequate time should be spent on isolating soil-dwelling animals in the laboratory. Furthermore, we recognise that the unequal number of sampling sites between the three regions (four localities in Hong Kong Island, two in the islands and fifteen in New Territories/Kowloon) might have also contributed to biased sampling effort. An expansion of collection sites in these other areas should be conducted in the future to generate a more complete and thorough survey. There is no doubt that the approach adopted in this study still has some technical flaws, but this study has raised public awareness and potentially opens up opportunities for the general public to engage in scientific research in the future.

In this study, we have only collected and identified 23 morphospecies. Comparing to the mentioned local public forum, HKWildlife.Net, some orders were not covered in this study, including Platydesmida, Siphonophorida and Chordeumatida. In addition, Cawjeekelia pallida (Golovatch 2011) and Polydesmus liber (Golovatch 1991) that were found in the local soil previously were not discovered in the current study. We believe that this is due to the biased sampling effort. Since most of the participating citizens are high school students and teachers, they have been focusing on the urban soil habitats that are within their own vicinity, while some millipedes, like T. corallinus and A. tonginus, are more commonly found in this project. This might be due to the reasons that these millipedes are more adapted to the urbanised environment as synanthropic species. We believed that more field surveys should be performed in the non-urbanised areas to broaden the sampling habitats in future studies.

Another interesting finding in this study is an unexpected discovery of Monographis queenslandica and Alloproctoides remyi in Hong Kong, which are the only two polyxenids identified locally. From the biogeographic perspective, it seems puzzling that with seven species of Monographis identified to date from Vietnam and two others in southern China, that the Monographis species identified in Hong Kong is the one that is found in Australia. We propose that there might be possibilities that the species was introduced to Hong Kong many decades ago in soil products or goods from Queensland, Australia or vice versa. The other species, A. remyi, was first identified from Reunion and more recently from Mauritius. A number of polyxenids from the family Lophoproctidae are also quite widespread in their distributions, so this might also be the case for A. remyi.

This study differs from most conventional scientific studies being carried out in Hong Kong, which were mainly carried out by either government, non-governmental organisations or academics in universities alone. Utilising a citizen science approach through creating a "big community" (a consortium; Suppl. material 6) in revealing the biodiversity, it also serves the purpose of “killing two birds with one stone” which could also educate the public and raise awareness on the use of basic science techniques in understanding local biodiversity. Some secondary school students had started millipede cultures in their own schools and there was one secondary school using the millipede breeding model to participate in a science and technology competition.

In terms of dissemination of the study, in addition to traditional academic output including this manuscript, sharing sessions by secondary school students were also carried out (online due to the COVID-19 pandemic). We have also summarised the findings and made online videos to a publicly-available online database/platform (http://biodiversity.sls.cuhk.edu.hk/millipedes) and designed a postcard (Fig. 6) which was distributed to participants as a souvenir. This study demonstrates a success in uniting local institutes and high schools and performing scientific research together with research teams in universities. We hypothesise that such bi-directional interaction could strengthen the bonding and engagement of participants involved in this citizen science project.

Figure 6.  

Designed postcard of millipedes identified in this study that was distributed to the participants. A Front; B Back.

This study has demonstrated a clear success in surveying the soil biodiversity in Hong Kong and opens up a possibility to carry out similar large-scale surveying via citizen science, together with taxonomic experts and researchers from universities. It is envisioned that the framework established in this study can also be adopted to reveal the biodiversity in other habitats in this region.