Over the last two decades there has been a marked increase in the intensity and frequency of proliferations of the benthic, mat-forming cyanobacterial genus Phormidium in rivers worldwide. This has raised concerns because species of Phormidium are known to produce a variety of cyanotoxins, including; anatoxin-a, homoanatoxin-a, dihydro-anatoxin-a and dihydro-homoanatoxin-a. Phormidium has already been implicated in animal toxicosis events in France, the Netherlands, Scotland, Switzerland and the United States of America. In New Zealand, there have been over 100 dog deaths attributed to ingestion of Phormidium since the early 2000’s. Concern over increasing Phormidium proliferations has led to a series of investigations into the underlying cause, but to date there is a limited understanding of the physicochemical factors which enhance growth. Previous observational studies have allowed inferences to be drawn that nutrients and river flow may play roles in determining Phormidium cover, however experimental evidence to test these inferences was limited. To address this gap, I have undertaken a series of observations and experiments investigating the relationship between Phormidium growth, nutrients and water velocity. Initially, I explored correlations between physicochemical variables and Phormidium cover and toxin concentrations in the Canterbury region (New Zealand), using an observational survey with high temporal resolution. A series of experiments followed that used novel field and mesocosm-based methods that allowed an in-depth analysis of how nutrients and flow affect different phases of the accrual cycle of Phormidium mats. These provided new insights into conditions under which proliferations are likely.

The observational-based field study was undertaken in eight rivers and involved weekly sampling over eight months. A variety of physicochemical variables including nutrients, dissolved metals, substrate size and stability, conductivity and temperature was measured as well as benthic Phormidium coverage and the associated toxin concentrations. Both Phormidium cover and anatoxin concentrations were highly variable spatially and temporally. Phormidium proliferations were documented under a wide range of water-column nutrient concentrations, including nitrogen concentrations suggested too low to support bloom formation in previous studies (dissolved inorganic nitrogen (DIN): <0.02 mg L-1) and also at low dissolved reactive phosphorus (DRP) concentrations (0.006 mg L-1). This survey used generalised additive mixed models to identify variables associated with Phormidium cover. High Phormidium cover was correlated with increasing conductivity and decreasing river flow. However, despite including a wide range of environmental and water quality variables, site was identified as an important factor in predicting cover, suggesting that a site-specific, timeindependent factor not included in analyses influences Phormidium cover. Higher anatoxin concentrations were measured between November and February and when DRP was less than 0.02 mg L-1. However, mats are a mixture of non-toxic and toxic genotypes and information of the relative abundance of these would allow more conclusive inferences to be drawn on influence of physicochemical factors on toxin concentrations.

Previous research has shown that Phormidium mats may access sediment-bound phosphorus trapped within mat matrices, allowing them to grow partially independent of river water nutrient supply, and the occurrence of proliferations at low DIN concentrations, led to the development of the hypothesis that nitrogen-fixing bacteria are present in Phormidium mats and could contribute to the nitrogen requirements for growth. I investigated this hypothesis using molecular techniques (i.e., high-throughput sequencing) and showed that 16 operational taxonomic units containing genes for nitrogen-fixation were present in mats, and that their diversity increased with increasing water-column nitrogen concentrations, rather than decreased as hypothesized. The potential for nitrogen fixation within mats may therefore explain a weak dependence on water-column DIN, though the rates of fixation relative to nitrogen demand were not investigated.

A stream-side mesocosm experiment was then conducted to elucidate links between Phormidium accrual and velocity and nitrate. Two velocity treatments; slow (0.1 m s-1) and fast (0.2 m s-1), were crossed with three nitrate treatments; ambient (0.02 mg L-1), medium (0.1 mg L-1) and high (0.4 mg L-1) and growth as areal expansion, biovolume, chlorophyll a and phycoerythrin was followed for 16 days. Due to the random nature of Phormidium colonisation in mesocosms, it was standardised by inoculating clean cobbles with a constant volume of homogenised Phormidium mats. High velocity resulted in significantly higher Phormidium biomass by the end of experiments, but did not affect expansion rates. Patches in slow velocity treatments detached significantly earlier compared to patches in high velocity treatments, due to autogenic detachment. Nitrate concentrations had no effect on Phormidium biovolume, chlorophyll a concentrations or mat expansion, although under high velocity and high nitrate treatments an increase in phycoerythrin concentrations over other treatments was evident.

In order to further investigate the effects of site and velocity on Phormidium accrual, I then undertook an in-stream experiment across a velocity gradient in three south Canterbury rivers, again using pre-inoculated cobbles to eliminate colonisation effects. Phormidium expansion rates and biomass (chlorophyll a and phycoerythrin concentrations and biovolumes) were measured, along with near-bed velocity, nutrients, macroinvertebrate communities and shear stress. Velocity determined mat morphology, expansion rate and patch longevity. Patches in low-medium velocity (i.e., pools and runs) habitats spread relatively quickly laterally, at least initially, whereas mats from the high-velocity riffles expanded at a slower rate, but increased in thickness. The optimal near-bed velocity for Phormidium rate of accrual, as patch size, was 0.25 to 0.45 m s-1, which in these rivers represented run habitats. Patches in riffles, though expanding more slowly, persisted throughout the experiment and this persistence may explain the tendency for riffles to have higher Phormidium cover. Patches from pools expanded rapidly but were also removed quickly, presumably due to high grazing pressure. Under similar velocities, differences in accrual were still evident among sites, which was partially attributed to differences in shear stress and macroinvertebrate communities.

Lastly, a synthesis of new knowledge gained through this thesis, together with recently published literature is presented, which develops a new paradigm for understanding the dynamics of Phormidium proliferation that is based on the factors influencing various phases of the accrual cycle. Of particular importance is the complex interactions between accrual, persistence and velocity, and perhaps nutrients. My research suggests that a nuanced interpretation of what governs Phormidium colonisation, growth and persistence needs to be developed that is sensitive to the demands of different phases of the accrual cycle. Future directions for research are also suggested, which aim to build upon the research presented in this thesis and further advance understanding of the physicochemical drivers of Phormidium proliferations.