Microbial micropatches within microbial hotspots

Abstract

The spatial distributions of organism abundance and diversity are often heterogeneous. This includes the sub-centimetre distributions of microbes, which have 'hotspots' of high abundance, and 'coldspots' of low abundance. Previously we showed that 300 μl abundance hotspots, coldspots and background regions were distinct at all taxonomic levels. Here we build on these results by showing taxonomic micropatches within these 300 μl microscale hotspots, coldspots and background regions at the 1 μl scale. This heterogeneity among 1 μl subsamples was driven by heightened abundance of specific genera. The micropatches were most pronounced within hotspots. Micropatches were dominated by Pseudomonas, Bacteroides, Parasporobacterium and Lachnospiraceae incertae sedis, with Pseudomonas and Bacteroides being responsible for a shift in the most dominant genera in individual hotspot subsamples, representing up to 80.6% and 47.3% average abundance, respectively. The presence of these micropatches implies the ability these groups have to create, establish themselves in, or exploit heterogeneous microenvironments. These genera are often particle-associated, from which we infer that these micropatches are evidence for sub-millimetre aggregates and the aquatic polymer matrix. These findings support the emerging paradigm that the microscale distributions of planktonic microbes are numerically and taxonomically heterogeneous at scales of millimetres and less. We show that microscale microbial hotspots have internal structure within which specific local nutrient exchanges and cellular interactions might occur.

Fig 1. Representative two-dimensional contour plot of total prokaryotic abundance showing hotspots, coldspots and background.

Circles indicate sample location and corresponding label. Faint gridlines indicate sampling intervals. Colour intensity scale in cells ml^-1. Two-dimensional contour plots were created via Surfer 10 (Golden Software, Inc.). Contour plots used a conservative minimum contour interval value of ≥ 1000 events μl^-1, which was higher than the maximum flow cytometric error of < 24 events μl^-1.

Fig 2. Rank abundance plot showing breaks and fitted distributions for hotspots (power), background (linear) and coldspots (power).

Fig 3

Genera profiles of (A) hotspot, (B) coldspot and (C) background subsamples. For clarity, only genera with average abundances ≥ 2% are shown. *incertae sedis. OTUs and associated abundance percentages determined via the RDP Classifier within the UPARSE pipeline [37, 41].

Fig 4

Metric multidimensional scaling (MDS) analysis using Bray-Curtis dissimilarity (A) and Sorensen distance metric (B) of subsamples from hotspot H1 (pink diamonds), H2 (grey crosses) and H3 (light blue circles), coldspot C1 (green squares), C2 (dark blue triangles) and C3 (red triangles), and background B1 (orange triangles), B2 (green crosses) and B3 (purple stars). Metric MDS ordination employed 500 bootstrap averages of the centroid of each sample to show where 95% of the centroid averages lie within multivariate space.

Fig 5. Violin plots of square root-transformed dAICc values for MLE fitted genus abundance distributions across all 60 samples.

A value of 0 indicates the best-fitting distribution; higher dAICc values indicate poorer fits.

Fig 6

Representative power laws from sample genus abundance distributions for (A) hotspots (B) background and (C) coldspots.