Microbial community assembly is a complex process driven by both deterministic (niche-based, selection-driven) and stochastic (random birth, death, dispersal) processes. Understanding the relative importance of these processes is crucial for managing microbial communities in engineered systems, particularly biofilms, which are prevalent in various applications, including wastewater treatment. Biofilm-based technologies offer advantages like reduced footprint and sludge production compared to suspended systems, and the enhanced microbial retention allows slow-growing organisms to thrive, potentially improving functionality. However, the ecological processes involved in biofilm assembly remain unclear, with studies showing varied conclusions regarding the dominance of deterministic or stochastic processes. Biofilm thickness, an emergent property influenced by hydrodynamic forces and substrate loading, is a potentially important factor affecting community assembly. This study hypothesized that stochastic processes would play a significant role in biofilm assembly and that selection pressures would vary with biofilm thickness, with intense competition expected in thin biofilms and increased variable selection in thicker ones due to environmental heterogeneity. The researchers aimed to examine the relative importance of deterministic and stochastic processes in biofilm community assembly across a range of thicknesses (50-500 µm), using Sloan's neutral community model and a null-modelling approach to analyze community diversity.

Existing research on microbial community assembly in engineered environments demonstrates a combined role for stochastic and deterministic processes. Studies on water filters, granular biofilm reactors, and microbial electrolysis cells highlight the context-dependent interplay of these processes. However, the factors determining their relative importance remain poorly understood. Several studies focusing on biofilms have yielded conflicting results, with some suggesting deterministic assembly dominance in mature stream biofilms while others emphasizing the role of stochasticity in systems like microbial electrolysis cells. This variation suggests that factors beyond biofilm formation, such as additional selective pressures, can confound the identification of assembly processes solely related to biofilm development. While the impact of biofilm thickness on biological treatment processes is widely discussed in modeling, fewer studies have investigated its role as a controllable parameter influencing community assembly. A previous study comparing two biofilm thicknesses emphasized deterministic processes but did not explicitly consider stochasticity. This current study addresses the gap by investigating the impact of biofilm thickness on both stochastic and deterministic processes in a more comprehensive and nuanced manner.

The study employed AnoxK™ Z-carriers in two continuous-flow nitrifying MBBRs to control biofilm thickness. Carriers with five grid heights (50, 200, 300, 400, and 500 µm) dictated maximum biofilm thickness. Reactors were run for approximately 350 days, fed with wastewater effluent supplemented with nutrients. Reactor 1 (3 l) contained various thicknesses, and reactor 2 (1.5 l) contained only 50 µm biofilms. Loading rates were similar for both reactors. Environmental parameters (temperature, pH, DO, HRT) were kept constant. DNA extraction, qPCR, and amplicon sequencing targeting the V3-V4 region of the 16S rRNA gene were performed on biofilm and influent samples at different timepoints. Bioinformatics involved quality control, merging of paired ends, error correction in DADA2, and taxonomic classification with SILVA. Data analysis in R used phyloseq, phangorn, vegan, and other packages. Neutral community modeling (Sloan's model) assessed the contribution of stochastic processes by comparing biofilm community structures with the influent community. Phylogenetic diversity analyses (Faith's PD, βMNTD, βMPD) and null models (999 runs) were used to examine deterministic processes. βNTI and βNRI were calculated to characterize phylogenetic structure relative to the null distribution. DPCoA was employed for ordination and identification of taxa driving differences. qPCR was used to quantify total bacterial cell numbers, allowing calculation of migration rates.

Neutral community modeling revealed that biofilms 200 µm and thicker had a better fit than 50 µm biofilms. Approximately half of the members of thicker biofilms could be explained by stochastic processes, comprising up to 75% of relative abundance, highlighting the significant role of stochastic processes in their assembly. However, habitat filtering was observed in all biofilms, particularly enriching nitrite-oxidizing Nitrospira spp. This enrichment was most pronounced in the thinner (50 µm) biofilms. The analysis of phylogenetic beta-diversity (βNTI and βNRI) further supported the significant role of habitat filtering across all biofilms. Notably, 50 µm biofilms exhibited greater clustering than expected by chance, suggesting homogeneous selection driven by competitive exclusion and physical forces. Thicker biofilms displayed phylogenetic overdispersion, potentially due to variable selection between replicates or drift. DPCoA revealed a clear separation between 50 µm and thicker biofilms, driven by the enrichment of Nitrospira spp. in thicker biofilms. Migration rates, calculated using the neutral model and qPCR data, were low across all biofilm thicknesses, indicating that dispersal played a minor role in assembly. The low migration rate and the high variability in community structure between biofilms of the same thickness revealed that substantial changes in community structure occurred over time even though biofilms were sampled once nitrogen removal had reached steady state.

The findings demonstrate a combined role for stochastic and deterministic processes in MBBR biofilm community assembly, with biofilm thickness significantly influencing their relative importance. Stochastic processes, primarily drift, explain a substantial portion of community composition in thicker biofilms, indicating that the biofilm lifestyle itself can be a relatively stochastic process. However, habitat filtering, leading to the enrichment of Nitrospira spp., was a consistently observed deterministic process across all biofilm thicknesses. The difference in assembly processes between thin and thicker biofilms can be attributed to varying selection pressures. In thin biofilms, homogeneous conditions and intense competition lead to the selection of closely related organisms, whereas thicker biofilms experience heterogeneous conditions resulting in variable selection or drift contributing to community dissimilarity among replicates. The low migration rates suggest that dispersal plays a minor role in shaping community structure in mature biofilms, especially those subjected to shear forces. These results challenge previous studies attributing community differences solely to deterministic processes, providing a more nuanced understanding of biofilm assembly dynamics. The observed differences in community assembly processes translate to functional differences in nitrification efficiency and micropollutant removal, with thinner biofilms demonstrating high nitrification rates and thicker biofilms exhibiting improved micropollutant transformation.

This study demonstrates that biofilm thickness significantly affects the relative importance of stochastic and deterministic processes in microbial community assembly within MBBRs. Stochasticity, mainly through drift, dominates community composition in thicker biofilms (≥200 µm). Habitat filtering, particularly enriching Nitrospira spp., acts as a deterministic force across all thicknesses. Thin biofilms (50 µm) exhibit homogeneous selection, while thicker biofilms show variable selection or drift effects. The results highlight the importance of considering biofilm thickness in managing microbial communities. Future research should focus on long-term biofilm succession studies to better understand community dynamics and test neutral community models with well-defined source communities and improved estimations of migration rates by incorporating effective community sizes.

A limitation of the experimental design was the use of separate reactors for the 50 µm biofilms and thicker biofilms, which started at different times. While efforts were made to control environmental factors and loading rates equally, the different reactor setup could have introduced some confounding factors. Although the influent microbial community composition was consistent, sampling at earlier timepoints during biofilm establishment, before steady state, could have revealed potentially different migration rates. Further studies with a more uniform setup are recommended to strengthen these conclusions. Finally, the study focused on a specific type of wastewater and MBBR setup, and the findings may not be universally applicable to all biofilm systems.