Microbiome composition of larval Zeugodacus cucurbitae
Three out of 27 phyla dominated the larval microbiomes: Proteobacteria, Firmicutes and Bacteroidetes (S7). In total, 122 orders, 205 families and 527 genera were identified in the larval microbiomes. The most abundant genera were Enterobacter (25%), Leuconostoc(9%), Acinetobacter (7.21%), Lactococcus (6.6%),Gluconobacter (6.15%) and Empedobacter (5.9%).
Across all host plants we identified seven core genera :Acinetobacter, Enterobacter, Klebsiella, Paenibacillus, Pseudomonas, Stenotrophomonas and Sphingobacterium .
Larval microbiomes showed a high ASV richness (mean = 145.36, SD = 75.56) but low phylogenetic diversity (mean = 13.51, SD = 4.58). Despite their high ASV richness, larval microbiomes were dominated by very few ASV as the ISI had a mean of 7.97 (SD = 6.21). The log transformed ISI did not show consistent patterns across sites or host plants (S6- S7). There was a significant interaction between host plant and site in the fourth root transformed ACE, the log transformed ISI and the fourth root transformed FPD (Table 1). The α diversity indices did not show consistent patterns across sites and host plants. ASV richness as estimated by ACE and FPD showed significantly higher values for Co in M, but not in BP. Conversely evenness showed significantly higher values for So in BP, but not in M (SI6-7).
PERMANOVA of the Generalized and Unweighted Unifrac distances revealed a significant interaction for β diversity between host plant and site (Table 2). However, pairwise comparisons did not provide additional information as all were significant (S8).
The PCoA on presence/absence data (Unweighted Unifrac distances) explains 46.15% of variation and suggests similarity between larvae feeding on different host plants (Fig. 1), however this pattern is not recovered in the PCoA on weighted abundancies (Generalized Unifrac distances, 59.03% of variation).
Several microbial genera showed a significant differential abundance between larvae raised on different host plants (S9 and Fig. 2-3).Ketogulonicigenium , Devosia, Salmonella, Allorhizobium and three possible unidentified genera from the family Enterobacteriaceae genera were more abundant on Co than on Cu and So. Most of the genera with higher abundance in So larvae belonged to the Order Enterobacterales, with Rahnella being detected in So larvae only. Compared to Cu larvae, larvae on So also showed enrichment in a few genera belonging to the Acetobacteriaceae, the class Bacilli and in the genus Pseudomonas . Larvae raised on Cu showed higher abundance for genera belonging to the bacterial families Clostridia and Negativicutes from the phylum Firmicutes and for the genusLeucobacter . An additional three genera of the Clostridia were more abundant in Cu than in Co larvae. In contrast, taxa belonging to Acetobacteraceae and to genus Pseudomonas were less abundant in larvae raised on Cu compared to larvae from other host plants.
Between sites, only the genus Comamonas showed a pattern of differentiation, as its relative abundance was 8.5x lower in larvae from M than in larvae from BP. Within hosts, more bacterial genera showed differential abundance between M and BP (S12).
Predictions from Tax4Fun2 suggested that bacteria might be involved in a variety of KEGG level 2 pathways important for host-symbiont interactions. The most abundant KEGG pathways include carbohydrate metabolism (18.8% of pathways), amino acid metabolism (11.8%), membrane transport (14.9%), cellular community (8.3%), cofactor and vitamin metabolism (6.1%), energy metabolism (6.3%), lipid metabolism (4.3%), nucleotide metabolism (4%) and xenobiotics degradation and metabolism (3.8%). Together these pathways involved over 76% of the predicted metabolic pathways (Fig. 4). Tax4Fun2 also identified some pathways associated with plant-pathogen interactions at low abundance (0.14%) and antibiotic resistance (2.5%).
Based upon the predictions of Tax4Fun2, there appear to be functional differences between larval and adult microbiomes. Fig. 4 shows that, overall, adults are more enriched in pathways related to carbohydrate metabolism and membrane transport. In contrast, amino acid metabolism appears to be more enriched in larvae than in adults.
There was considerable variation between pooled larval samples in the fraction of ASV’s and sequences that could be used by Tax4Fun2 to make these predictions. Between 51.3% and 95.5% of all ASV’s and between 63% and 99.9% of sequences in samples were used in metagenomic predictions.