Sommeria-Klein Guilhem, Zinger Lucie, Taberlet Pierre, Coissac Eric, Chave Jérôme
Université Toulouse 3 Paul Sabatier, CNRS, UMR 5174 Laboratoire Evolution et Diversité Biologique, F-31062 Toulouse, France.
Université Grenoble Alpes, CNRS, UMR 5553 Laboratoire d'Ecologie Alpine, F-38000 Grenoble, France.
Sci Rep. 2016 Oct 20;6:35644. doi: 10.1038/srep35644.
The DNA present in the environment is a unique and increasingly exploited source of information for conducting fast and standardized biodiversity assessments for any type of organisms. The datasets resulting from these surveys are however rarely compared to the quantitative predictions of biodiversity models. In this study, we simulate neutral taxa-abundance datasets, and artificially noise them by simulating noise terms typical of DNA-based biodiversity surveys. The resulting noised taxa abundances are used to assess whether the two parameters of Hubbell's neutral theory of biodiversity can still be estimated. We find that parameters can be inferred provided that PCR noise on taxa abundances does not exceed a certain threshold. However, inference is seriously biased by the presence of artifactual taxa. The uneven contribution of organisms to environmental DNA owing to size differences and barcode copy number variability does not impede neutral parameter inference, provided that the number of sequence reads used for inference is smaller than the number of effectively sampled individuals. Hence, estimating neutral parameters from DNA-based taxa abundance patterns is possible but requires some caution. In studies that include empirical noise assessments, our comprehensive simulation benchmark provides objective criteria to evaluate the robustness of neutral parameter inference.
环境中的DNA是一种独特且越来越多地被利用的信息来源，可用于对任何类型的生物体进行快速且标准化的生物多样性评估。然而，这些调查产生的数据集很少与生物多样性模型的定量预测进行比较。在本研究中，我们模拟了中性分类单元丰度数据集，并通过模拟基于DNA的生物多样性调查中典型的噪声项对其进行人工噪声处理。由此产生的带噪声的分类单元丰度用于评估哈贝尔中性生物多样性理论的两个参数是否仍可估计。我们发现，只要分类单元丰度上的PCR噪声不超过某个阈值，参数就可以推断出来。然而，人为分类单元的存在会严重影响推断结果。由于大小差异和条形码拷贝数变异性，生物体对环境DNA的贡献不均一，但只要用于推断的序列读数数量小于有效采样个体的数量，就不会妨碍中性参数的推断。因此，从基于DNA的分类单元丰度模式估计中性参数是可能的，但需要谨慎。在包括实证噪声评估的研究中，我们全面的模拟基准提供了评估中性参数推断稳健性的客观标准。
