Fig. 5

Bacterial communities of sponge developmental stages. A Average phyla-level taxonomic profile for the developmental stages of 35 sponge species (33 Demospongiae and 2 Homoscleromorpha; 14 HMA species and 21 LMA species) [96, 114, 116, 117]. This community includes 7794 unique taxa (ASVs) from 40 phyla and were predominantly composed of Proteobacteria (56.3%), Chloroflexi (9.8%), Acidobacteria (7.6%), Actinobacteria (5.2%), Cyanobacteria (4.0%), and Firmicutes (3.6%). B Diversity of the bacterial communities associated with sponge developmental stages was assessed by calculating the total ASVs, Faith’s phylogenetic diversity, and McIntosh dominance. The gray dots represent an average value for each of the 35 species and the black dot is the average for all species (± 95% confidence intervals). HMA and LMA sponges associated with bacterial communities that were comparably similar in total taxa and phylogenetic diversity and LMA sponges were dominated by a few bacterial taxa and HMA sponges were more evenly distributed. C Principal coordinates analysis depicting community relatedness of microbiome composition (based on weighted UniFrac values) and show that the developmental stages of sponges tend to associate with a species-specific microbiota while HMA and LMA sponges associate bacterial communities that are compositionally distinct, providing the first demonstration of the HMA-LMA dichotomy in the developmental stages. D Dendrogram of microbiome relatedness for the 25 sponge species with two or more samples, as based on and simplified from the principal coordinates analysis (C). Each sample in C and D is represented as a dot. All species are color coded and sponge lifestyle is coded by shape (HMA sponges are represented by squared and LMA sponges are represented by circles). To analyze these amplicon data, raw reads (along with quality information) from four studies [96, 114, 116, 117] were downloaded directly from the Sequence Read Archive of the National Center for Biotechnology Information (Table S2) and were imported into QIIME 2 (v.2019.1 [118]). Primers were trimmed, sequences were filtered by quality score, and each sample was denoised using Deblur [119]. QIIME 2-generated “features” were analyzed as amplicon sequence variants (ASVs [120]) and were assigned taxonomy using SILVA (v.132 [121]). The few sequences matching to Archaea and mitochondria as well as singletons were removed. Samples with less than 1000 reads were discarded, and the filtered data table was rarified to 1675 sequences (Table S10)