Abstract:

Objectve: How to accurately estimate species richness in some specific area is a basic scientific question in ecological research. In this study, a 60 hm² large-scale forest dynamic plot in Jianfengling National Nature Reserve of Hainan was selected as the typical mountain rain forest for quantification of soil microbial diversity. Species richness of soil bacteria and fungi were estimated using high-throughput sequencing and statistical inference estimation methods. Method: A total of 500 soil samples were collected from 0-10 cm soil layers in each 40 m×30 m quadrat in this 60 hm² large-scale plot. Illumina MiSeq sequencing platform was used to measure the bacterial 16S rRNA and fungal ITS2 (internal transcribed spacer 2) rRNA gene sequences of the soil samples. Parametric estimators SAC (species accumulative curve) and SAD (species abundance distribution), and non-parametric estimators Chao2, ICE (Incidence-based Coverage Estimator) and Jackknife1 were used to estimate the OTU richness of soil microbial community in all samples, and their effectiveness was evaluated. Based on the abundance and frequency of OTU in all samples, the sample coverage was analyzed by using R package iNEXT (iNterpolation and EXTrapolation of Hill number). Result: From the 500 soil samples, the number of high-quality DNA sequences belonging to soil bacteria and fungi was 14 000 317 and 7 656 130, respectively. These sequence data were identified as 42 873 and 22 923 OTUs for soil bacterial and fungal communities, respectively. Comparing the estimation results based on the five parameters and nonparametric estimators, it was found that the difference among the five estimators was small, and all of them slight overestimated the number of observed species. Non-parametric Chao2 estimator was the best estimator to accurately estimate the soil bacterial and fungal OTU richness, with 42 828 bacterial OTUs and 23 137 fungal OTUs in the 60 hm² plot. The result of iNEXT analysis showed that the number of obtained sequences and the collected soil samples in the study covered more than 99% of the microbial populations in the 60 hm² plot. In particular, abundance-based sample coverage reached 100%. Conclusion: Non-parametric Chao2 was the best estimator of all five estimators, which showed a lower extrapolation error. The iNEXT analysis showed that at least 281 soil samples and 742 767 DNA sequences were required to cover 99% of the soil bacteria OTU richness, while at least 386 soil samples and 383 189 DNA sequences were needed to estimate the OTU richness of soil fungal communities. In sum, the present study provides a basis for investigating the soil microbial diversity in a 60 hm² large-scale plot in the tropical mountain rain forest of Jianfengling, Hainan Island and provides references for estimating methods, sample volumes, and sequencing depth for better estimation of microbial diversity.

Key words: tropical mountain rain forest, soil microbe, high-throughput sequencing, species richness