Human population pressures and activities pose unprecedented challenges to water resources in urban environments. However, standard methods of assessing microbial water quality have relied on the same cultured organisms for decades. We show that there is a conserved microbial assemblage in untreated sewage that can be exploited to improve global sewage surveillance. Among harbour and coastal water samples from 18 cities across 5 continents (n = 442), nearly half had evidence of sewage contamination using two human faecal bacteria as molecular indicators. In contrast, conventional measures using cultured Escherichiacoli or enterococci only exceeded water quality limits in similar to 18% of samples, with less than half of these demonstrating sewage indicators. Contaminated locations also displayed a signature characteristic of microorganisms mainly derived from sewer infrastructure. Given the human health risk, loss of ecosystem services and economic costs associated with contaminated coastal waters, molecular approaches could provide more reliable information on sewage contamination of urban waterways.

Aquatic microbial ecosystems are increasingly under threat from human activities, highlighting the need to for the development and application of biomonitoring tools that can identify anthropogenically induced stress across a wide range of environments. To date, microbial biomonitoring has generally focussed on community composition and univariate endpoints, which do not provide discrete information about how species both interact with each other and as a collective. To address this, co-occurrence networks are being increasingly used to complement traditional community metrics. Co-occurrence network analysis is a quantitative analytical tool that examines the interactions between nodes (e.g. taxa) and their strengths. This information can be integrated and visualised as a network, whose characteristics and topological structures can be quantified. To date, co-occurrence network analysis has rarely been applied to aquatic systems. Here we explore the potential of co-occurrence networks as a biomonitoring tool in aquatic environments, demonstrating its capacity to provide a more comprehensive view of how microbial, notably bacterial, communities may be altered by human activities. We examine the key attributes of networks and providence evidence of how these may change as a response to disturbances while also highlighting some of the challenges associated with making the approach routine.

The Hawkesbury-Nepean River (HNR) is the largest catchment in the Sydney region and is undergoing unprecedented population growth. The HNR system receives a mix of anthropogenic inputs such as treated sewage, stormwater and agricultural runoff. Combined, these can diminish the ecological system health and pose potential concerns to human health. Of particular concern are inputs of untreated sewage, that can occur due to a range of different reasons including illegal point source discharges, failure of the sewerage network, and overloading of wastewater treatment plants during storm events. Here, we present findings of an intensive assessment across the HNR catchment where we used a weight-of-evidence (WOE) approach to identify untreated sewage contamination in surface waters against the background of treated effluent and diffuse inputs during post high flow conditions. Total nitrogen and phosphorus concentrations were used to assess treated effluent and diffuse inputs, and microbial analysis, including both culture-based traditional methods for E. coli and enterococci and qPCR analysis of Bacteroides and Lachnospiraceae, were used to assess raw sewage contamination. Despite a background of diffuse inputs from recent high flow events and the influence of treated wastewater, we found no gradient of faecal contamination along the HNR system or its tributaries. We observed two sites with evidence of untreated sewage contamination, where the human markers Bacteroides and Lachnospiraceae qPCR copy numbers were high. The biological and chemical approaches suggested these latter two hotspots originate from an industrial runoff source and possibly from a dry weather sewage leak. Our findings demonstrate the potential of a WOE approach in the assessment of human faecal signal in an urban river that can also pinpoint small sources of contamination as a strategy that can reshape the way monitoring is performed and the chemical end-points chosen to provide pertinent information on the potential risks to aquatic system health.