Figs. 3B and 4B appear identical in the online publication of this article. The corrected Fig. 4B is shown below. [Figure presented] Additional information for the 18S rDNA high throughput sequences that affiliated to non-fungal Eukaryota In recent years, the use of high-throughput sequencing technologies with the 16S rRNA gene for characterization of bacterial or archaeal communities has become more popular. However, the adoption of such techniques for eukaryotes has been relatively slow, despite their significance in both natural and engineered systems. This situation is primarily a result of the large variation in copy numbers of targeted genes among different species, and multiple hypervariable regions that are typically longer than early DNA sequencing platforms could handle (Bradley et al., 2016). In the article “Differences of Microbial Community on the wall paintings preserved in situ and ex situ of the Tiantishan Grottoes, China” published on the journal of International Biodeterioration & Biodegradation, we employed a typical fungal primer SSU-0817F, 5′-TTAGCATGGAATAARRAAG-3′ and SSU-1196R, 5′-TCTGGACCTG GTGAGTTTCC-3′ targeting the V5-V7 regions of the 18S rRNA genes. These primers were chosen as they flank a conserved region of the 18S gene that differs among fungal families, but they are similar enough to be aligned in multiple sequence alignments. This 18S region is often considered to be more conserved than the ITS regions; the use of 18S region may lower diversity estimates and make the obtained data to be at the family level (Kivlin et al., 2014). Thus, the universal primers 817F/1196R for fungi was used for the amplification and then MiSeq sequencing of the PCR products in our research. Of course, many reports show that it also has a slightly but inevitable weakness for successful detection of fungi. Our results show that the most of the reads belonged to the fungal phyla of Ascomycota (197,465, 90.45% for reads and relative abundance, respectively) and Basidiomycota (11,289, 5.17%). Meanwhile, very tiny fractions of reads were classified as non-fungal Eukaryota, but they are typically detected at relatively low abundances (ca. 2.73%); most of them were affiliated with the Vertebrata, Ciliophora, Arthropoda and Bicosoecida (Duan et al., 2018). Similar phenomenon has reported repeatedly in the past several years, for example, many sequences classified into Bacillariophyta, Dinophyceae, Dictyochophyceae, Rotifera, Ciliophora, Arthropoda, Cercozoa, Telonema and Prymnesiales using the 18S rDNA-DGGE method (Cunliffe and Murrell, 2010). Anderson et al. (2003) used the universal primers 817F/1196R, same as ours, retrieved several non-fungal sequences, such as two species of invertebrates of Crossodonthina koreana and Hypogastrura dolsana; Rhinosporidium seeberi, an aquatic protistan parasite; Diplolaimelloides meyli, a soil arthropod. Chen et al. (2016) also used this universal primers 817F/1196R for fungi and retrieved a tiny fraction of non-fungal sequence, classified into Cercozoa, Centrohelida, Choanomonada and Ciliophora. It is widely believed that the discrepancies in primer specificity may become more apparent over time as the primers are tested further, particularly with samples collected from diversely different environments containing a diversity of eukaryotic communities (Anderson et al., 2003). Although the primer pair 817F/1196R is lack of the desired specificity for characterizing fungal communities alone, it is still a useful tool for analyzing fungal communities in environmental samples, particularly for amplification of 18S rDNA sequences from several fungal phyla, such as Ascomycota, Basidiomycota, Chytridiomycota, and Zygomycota. In fact, the universal primers 817F/1196R was optimal for Illumina MiSeq PE300 Platform and widely used by mainstream sequencing providers. The high-throughput sequencing procedure will be producing a huge number of raw reads, and all of them must be deposited into the NCBI Sequence Read Archive (SRA) database. If elimination of the non-fungal sequences from the data pool, the relative abundance of each taxonomic member across the samples must be slightly changed. Therefore, it is suggested to keep the non-fungal Eukaryota sequences within the raw data.