Exploring freshwater stream bacterial communities as indicators of land use intensity.

BACKGROUND: Stream ecosystems comprise complex interactions among biological communities and their physicochemical surroundings, contributing to their overall ecological health. Despite this, many monitoring programs ignore changes in the bacterial communities that are the base of food webs in streams, often focusing on stream physicochemical assessments or macroinvertebrate community diversity instead. We used 16S rRNA gene sequencing to assess bacterial community compositions within 600 New Zealand stream biofilm samples from 204 sites within a 6-week period (February-March 2010). Sites were either dominated by indigenous forests, exotic plantation forests, horticulture, or pastoral grasslands in the upstream catchment. We sought to predict each site's catchment land use and environmental conditions based on the composition of the stream bacterial communities. RESULTS: Random forest modelling allowed us to use bacterial community composition to predict upstream catchment land use with 65% accuracy; urban sites were correctly assigned 90% of the time. Despite the variation inherent when sampling across a ~ 1000-km distance, bacterial community data could correctly differentiate undisturbed sites, grouped by their dominant environmental properties, with 75% accuracy. The positive correlations between actual values and those predicted by the models built using the stream biofilm bacterial data ranged from weak (average log N concentration in the stream water, R2 = 0.02) to strong (annual mean air temperature, R2 = 0.69). CONCLUSIONS: Freshwater bacterial community data provide useful insights into land use impacts on stream ecosystems; they may be used as an additional measure to screen stream catchment attributes.

A predictive model of recreational water quality based on adaptive synthetic sampling algorithms and machine learning.

Predicting recreational water quality is one of the most difficult tasks in water management with major implications for humans and society. Many data-driven models have been used to predict water quality indicators to allow a real time assessment of public health risk. This assessment is most commonly based on Faecal Indicator Bacteria (FIB), with the value of FIB compared with thresholds published in guidelines. However, FIB values usually tend to be unbalanced within water quality datasets, with small proportions of data exceeding guideline thresholds and far larger numbers that do not. This can be a limiting factor in the uptake of model predictions since, even if the overall accuracy is high, the sensitivity of the predictions can be low. To address this issue, this paper proposes an adaptive synthetic sampling algorithm (ADASYN) to generate synthetic above-threshold FIB instances and test the validity of the approach for the prediction of recreational water quality. The models in this paper are based on four machine learning techniques: k-mean nearest neighbour, boosting decision tree, support vector machine, and multi-layer perceptron artificial neural network and are applied to five different locations in Auckland, New Zealand. Aside from support vector machine, all models provide favourable predictions with relatively high sensitivity (around 75%) and overall accuracy (over 90%), indicating that both the compliant and exceedance conditions can be effectively predicted through the use of more sophisticated model training which involves artificial data. Considering the model accuracy and stability, boosting decision trees (BDT) and multi-layer perceptron artificial neural (MLP-ANN) network are the best two models and the multi-layer perceptron is the most efficient with the shortest computation time.

Following Rapoport's Rule: the geographic range and genome size of bacterial taxa decline at warmer latitudes.

We sought to test whether stream bacterial communities conform to Rapoport's Rule, a pattern commonly observed for plants and animals whereby taxa exhibit decreased latitudinal range sizes closer to the equator. Using a DNA sequencing approach, we explored the biogeography of biofilm bacterial communities in 204 streams across a ∼1000 km latitudinal gradient. The range sizes of bacterial taxa were strongly correlated with latitude, decreasing closer to the equator, which coincided with a greater than fivefold increase in bacterial taxonomic richness. The relative richness and range size of bacteria were associated with spatially correlated variation in temperature and rainfall. These patterns were observed despite enormous variability in catchment environmental characteristics. Similar results were obtained when restricting the same analyses to native forest catchments, thereby controlling for spatial biases in land use. We analysed genomic data from ∼500 taxa detected in this study, for which data were available and found that bacterial communities at cooler latitudes also tended to possess greater potential metabolic potential. Collectively, these data provide the first evidence of latitudinal variation in the range size distributions of freshwater bacteria, a trend which may be determined, in part, by a trade-off between bacterial genome size and local variation in climatic conditions.

Using biofilm as a novel approach to assess stormwater treatment efficacy.

Contaminants associated with stormwater are among the leading causes of water quality impairment in urban streams. Multiple device treatment systems are commonly installed with the aim of reducing contaminant loads within stormwater discharge. However, the in situ performance of such systems remains poorly understood. We investigated the efficacy of an advanced stormwater treatment system by monitoring biofilm associated metals and biofilm bacterial community composition at multiple locations through the treatment system (which included rain gardens, grassy swales, a stormwater filter and a wetland) and in the receiving stream above and below the stormwater discharge. Changes in bacterial community composition were assessed by Automated Ribosomal Intergenic Spacer Analysis (ARISA) and concentrations of biofilm associated metals monitored by Inductively Coupled Plasma-Mass Spectrometry (ICP-MS). Significant differences in bacterial community composition were detected throughout the stormwater network. Bacterial communities gradually changed towards a community more similar to that within the receiving stream and the discharge of treated stormwater had little effect on the composition of bacterial communities in the receiving stream, suggesting the effective conditioning of water quality by the treatment system. Concentrations of some biofilm-associated metals declined following sequential treatment, for example copper (73% reduction), zinc (48% reduction) and lead (46% reduction). In contrast, levels of arsenic, cadmium, chromium and nickel were not reduced by the treatment system. We demonstrate that biofilm bacterial community composition is a sensitive indicator of environmental changes within freshwater ecosystems and an efficient indicator to monitor water quality in enclosed stormwater networks where traditional biological indicators are not available.