Prediction of Minimum Postmortem Submersion Interval (PMSImin) Based on Eukaryotic Community Succession on Skeletal Remains Recovered from Aquatic Environments. PMSI Prediction using 18S rDNA during aquatic Decomposition

Many recent studies were conducted concerning bacterial succession in decomposing animal carrion in terrestrial system. However, not much is known about the microorganisms involved in aquatic decomposition. To the best of our knowledge, there are currently no published studies which estimate minimum postmortem submersion interval PMSImin based on eukaryotic community succession in aquatic system. The main goal of this study was to determine the eukaryotic community succession associated with porcine skeletal remains in lentic and lotic environments, and to derive a statistical model for PMSImin prediction. The locations chosen for this research were Henleys Lake in Crozet, VA and Virginia Commonwealth University’s (VCU’s) Rice Rivers Center on the James River. The distance between both sites was approximately 120 miles. At each site, fresh pig bones (rib N = 100, scapula N = 100) were placed in cages (10 x 10 inch2), attached to a floatation device, and submerged together with waterproof loggers, and a Yellow Springs, OH Sonde. Every 250 Accumulated Degree Days (ADD), one cage containing 5 rib and 5 scapula samples was collected over a total research period of 5200 ADDs (559 days) at Henleys Lake, and 4972 ADDs (294 days) at the James River. Water samples were also collected every 250 ADD and filtered using a cellulose membrane filtration system. DNA extraction was performed using the Invitrogen ChargeSwitch® gDNA Plant Kit Protocol. Variable region nine (V9) of the 18S rDNA was amplified and sequenced using a dual-index strategy on the MiSeq FGX sequencing platform. Sequenced data were quality controlled and analyzed via the MiSeq SOP in Mothur version 1.42.3 and in R v3.6.0. Hierarchical classification of good quality sequences was performed using SILVA119 reference database. A phylogenetic approach was utilized for α-and β-diversity estimation. For relative abundance and diversity estimations, sequences were subsampled at a threshold of 5048 reads (Henleys Lake) and 5287 reads (James River). Permutational multivariate analysis of variance (PERMANOVA) revealed a significant difference in eukaryotic community structure among sample types (p = 0.000999) and ADD (p ≤ 0.01099). Non-metric multidimensional scaling (NMDS) ordinations revealed distinct clustering of samples associated with each ADD. Arthropoda_unclassified was the most influential PMSI predictor taxon at Henleys Lake, while Eukaryota_unclassified was the most influential PMSI predictor taxon at the James River. Two random forest models were generated for rib and scapula at each site. At Henley’s lake, rib and scapula models predicted PMSImin with an error rate of ± 101 days (936.89 ADD) and ± 61 days (563.77 ADD) respectively. At the James River, rib and scapula models predicted PMSImin with an error rate of ± 32 days (547.44 ADD) and ± 31 days (527.08 ADD) respectively. Scapula was found to be a better predictor than Rib at both sites. Future directions include comparison of 18S rDNA data between lentic and lotic bodies of water. There is also potential for this 18S rDNA data to be combined with 16S rDNA data to increase accuracy of PMSImin prediction models. In conclusion, this study suggests that eukaryotic succession is capable of predicting long term PMSI in freshwater systems.

目前已有诸多研究针对陆地系统中腐解动物尸体的细菌演替展开，但针对水生腐解过程中涉及的微生物的相关认知仍较为匮乏。据我们所知，目前尚无公开研究基于水生系统中的真核生物群落演替来估算最小死后沉没间隔（minimum postmortem submersion interval, PMSImin）。
本研究的核心目标为明确静水（lentic）与流水（lotic）环境中猪骨骼遗骸相关的真核生物群落演替规律，并构建用于预测PMSImin的统计模型。本研究选取的试验地点分别为弗吉尼亚州克罗泽特的亨莱湖（Henleys Lake），以及詹姆斯河畔的弗吉尼亚联邦大学（Virginia Commonwealth University, VCU）赖斯河流中心（Rice Rivers Center），两地间距约120英里。
在每个试验地点，我们将新鲜猪骨（肋骨100份、肩胛骨100份）放置于10×10平方英寸的笼具中，并附着漂浮装置，与防水记录仪及俄亥俄州耶洛斯普林斯市（Yellow Springs, OH）生产的水质探头一同沉入水中。每累积度日数（Accumulated Degree Days, ADD）达到250时，即采集一个装有5份肋骨样本与5份肩胛骨样本的笼具；亨莱湖的总试验周期为5200 ADD（对应559天），詹姆斯河试验点的总试验周期为4972 ADD（对应294天）。同时每250 ADD采集一次水样，采用纤维素膜过滤系统进行过滤。
DNA提取采用Invitrogen ChargeSwitch® 基因组DNA植物提取试剂盒操作流程。通过双索引策略，针对18S核糖体DNA（18S rDNA）的可变区九（V9）进行扩增与测序，测序平台为MiSeq FGX测序平台。测序数据的质控与分析分别采用Mothur v1.42.3版本中的MiSeq标准操作流程（MiSeq SOP）以及R v3.6.0软件完成。使用SILVA119参考数据库对高质量序列进行层级分类。采用系统发育学方法估算α多样性与β多样性。在相对丰度与多样性估算环节，将亨莱湖样本的测序量抽平至5048条读段，詹姆斯河样本的测序量抽平至5287条读段。
置换多元方差分析（permutational multivariate analysis of variance, PERMANOVA）结果显示，不同样本类型（p=0.000999）与ADD梯度（p≤0.01099）下的真核生物群落结构均存在显著差异。非度量多维尺度分析（non-metric multidimensional scaling, NMDS）排序结果显示，不同ADD梯度对应的样本呈现出明显的聚类特征。在亨莱湖，未分类节肢动物门（Arthropoda_unclassified）是对PMSImin预测贡献最大的分类单元；而在詹姆斯河，未分类真核生物域（Eukaryota_unclassified）为最优预测分类单元。
本研究在每个试验地点分别针对肋骨与肩胛骨构建了随机森林模型。在亨莱湖，肋骨样本模型与肩胛骨样本模型预测PMSImin的误差分别为±101天（对应936.89 ADD）与±61天（对应563.77 ADD）。在詹姆斯河，肋骨样本模型与肩胛骨样本模型预测PMSImin的误差分别为±32天（对应547.44 ADD）与±31天（对应527.08 ADD）。在两个试验地点，肩胛骨样本的预测性能均优于肋骨样本。
未来研究可对比静水与流水水体中的18S rDNA数据，同时也可将18S rDNA数据与16S rDNA数据结合，以提升PMSImin预测模型的准确性。综上，本研究表明真核生物群落演替可用于预测淡水生态系统中长期的死后沉没间隔。