Antimicrobial resistance (AMR) increases the burden of infectious disease by slowing or preventing treatment, leading to worsened health outcomes. Hospitals and human communities are recognised sources of infections by resistant bacteria. This research broadens the search for and characterisation of infection vectors to water and mahinga kai. In Aotearoa New Zealand people are close to their water, using it for recreation or food gathering. Mahinga kai are traditional foods and resources of Māori and the places where they are harvested. These foods remain culturally and nutritionally important for both Aotearoa’s indigenous people and others.

Many mahinga kai species are aquatic. Harvesting and consumption of mahinga kai may therefore be a risk factor for exposure to the same bacteria that contaminate water, such as Escherichia coli. There is evidence that E. coli is not only a sentinel species for fecal contamination, but also for the presence of other antimicrobial resistant fecal coliforms, many of which cause disease. Moreover, up to half of antimicrobial resistant infections worldwide are caused by E. coli. There is limited data on AMR among E. coli in Aotearoa’s surface waters, and carriage by mahinga kai remains largely unassessed.

I performed observational and hypothesis driven experiments to study AMR in aquatic environments. I extended the work of myself and colleagues through two years of seasonal monitoring of Ōtākaro, an urban river. The data were used to develop sentinels that would allow local iwi, community groups and governing bodies to monitor AMR with the least input of resources. Most interestingly, the concentration of ampicillin resistant E. coli could be used to predict the likelihood of encountering E. coli resistant to either of two additional antibiotics, ciprofloxacin or chloramphenicol, which are biochemically unrelated antibiotics of clinical relevance. The use of ‘sentinel antibiotics’ could be a powerful way to increase monitoring capability.

Over the years I monitored Ōtākaro, a large number of antimicrobial resistant E. coli were analysed. These were phenotyped for antimicrobial susceptibility and conjugation, and genotyped through whole genome sequencing. The combination of phenortyping and

genotyping provided a comprehensive description of the types of drug resistances, their potential for spread through horizontal gene transfer (HGT) and underlying genetic mechanisms. I showed that plasmid-mediated HGT is likely a major contributor to AMR in these E. coli because resistance genes were often linked on conjugative plasmids. Long-read whole genome sequencing was used to assemble plasmids, demonstrating the association of AMR genes and genes associated with virulence and stress tolerance. These plasmids were often hosted by E. coli of sequence types known to be human pathogens, such as ST131 and ST1193. The presence of these sequence types in Ōtākaro may indicate underlying carriage by people in the community and poses a health risk for contact recreation.

Waitaha North Canterbury has long been a food basket for the tribe Ngāi Tahu, particularly Te Ngāi Tūāhuriri. At their request, I studied E. coli and antimicrobial resistant E. coli in wātākirihi (watercress; Nasturtium officinale) from rivers near Tuahiwi and in tuaki (cockles; Austrovenus stuchburyii) from Te Akaaka (Ashley River Estuary). Both kai species and the water in which they grow contained antimicrobial resistant E. coli. The concentration of E. coli in water was predictive of E. coli concentration on wātākirihi. Water quality monitoring data can therefore help wātākirihi harvesters avoid exposure to E. coli. Tuaki were sensitive environmental monitors and were able to concentrate E. coli up to 60 times higher than in surrounding water. This allowed me to detect AMR phenotypes inside tuaki that were not detectable in congruent water samples. Furthermore, the concentration of E. coli in water was not a good predictor of E. coli concentration in tuaki, making water samples insufficient for predicting food safety risk. I also present preliminary data on antimicrobial resistant E. coli in karoro (black-backed gulls; Laurus dominicus) feces and on eggs. Karoro eggs and feathers can be mahinga kai.

I then tested a series of hypothesis on the accumulation and fate of AMR phenotypes of E. coli in shellfish and in seawater. I did this using kuku (Greenshell mussels; Perna canaliculus) maintained in aquaria. E. coli were efficiently taken up by kuku, which became unsafe for consumption after 8-24 hours of exposure. However, there was no difference in the uptake and accumulation of a conjugative, multidrug resistant E. coli strain in kuku when compared

to a susceptible strain. Finally, I tested the hypothesis that, by concentrating bacteria, shellfish may promote the spread of AMR via conjugation.

My results provide baseline data on AMR in urban and rural rivers and in mahinga kai extracted from these waterways. I stress the importance of better measures to prevent fecal contamination to protect water and mahinga kai users.