Testing the intrinsic mechanisms driving the dynamics of Ross River Virus across Australia.

The mechanisms driving dynamics of many epidemiologically important mosquito-borne pathogens are complex, involving combinations of vector and host factors (e.g., species composition and life-history traits), and factors associated with transmission and reporting. Understanding which intrinsic mechanisms contribute most to observed disease dynamics is important, yet often poorly understood. Ross River virus (RRV) is Australia's most important mosquito-borne disease, with variable transmission dynamics across geographic regions. We used deterministic ordinary differential equation models to test mechanisms driving RRV dynamics across major epidemic centers in Brisbane, Darwin, Mandurah, Mildura, Gippsland, Renmark, Murray Bridge, and Coorong. We considered models with up to two vector species (Aedes vigilax, Culex annulirostris, Aedes camptorhynchus, Culex globocoxitus), two reservoir hosts (macropods, possums), seasonal transmission effects, and transmission parameters. We fit models against long-term RRV surveillance data (1991-2017) and used Akaike Information Criterion to select important mechanisms. The combination of two vector species, two reservoir hosts, and seasonal transmission effects explained RRV dynamics best across sites. Estimated vector-human transmission rate (average ß = 8.04x10-4per vector per day) was similar despite different dynamics. Models estimate 43% underreporting of RRV infections. Findings enhance understanding of RRV transmission mechanisms, provide disease parameter estimates which can be used to guide future research into public health improvements and offer a basis to evaluate mitigation practices.

Citizen Science Mosquito Surveillance by Ad Hoc Observation Using the iNaturalist Platform.

Citizen science mosquito surveillance has been growing in recent years due to both increasing concern about mosquito-borne disease and the increasing popularity of citizen science projects globally. Health authorities are recognising the potential importance of citizen science to expanding or enhancing traditional surveillance programs. Different programs have shown success in engaging communities to monitor species of medical importance through low-cost methods. The Mozzie Monitors project was established on iNaturalist-an open citizen science platform that allows participants to upload photos (i.e., observers) and assist identification (i.e., identifiers). This article describes the likelihood of citizen scientists submitting photos of mosquitoes, assesses user submission behaviour, and evaluates public health utility from these citizen science-derived data. From October 2018 to July 2021, the Mozzie Monitors project on iNaturalist received 2118 observations of 57 different species of mosquitoes across Australia. The number of observers in the system increased over time with more than 500 observers and 180 identifiers being active in the project since its establishment. Data showed species bias with large-bodied and colourful mosquitoes being over-represented. Analyses also indicate regional differentiation of mosquito fauna per state, seasonality of activity, and ecological information about mosquitoes. The iNaturalist citizen science platform also allows connectedness, facilitated communication and collaboration between overall users and expert entomologists, of value to medical entomology and mosquito management.

Using ecological variables to predict Ross River virus disease incidence in South Australia.

BACKGROUND: Ross River virus (RRV) disease is Australia's most widespread vector-borne disease causing significant public health concern. The aim of this study was to identify the ecological covariates of RRV risk and to develop epidemic forecasting models in a disease hotspot region of South Australia. METHODS: Seasonal autoregressive integrated moving average models were used to predict the incidence of RRV disease in the Riverland region of South Australia, an area known to have a high incidence of the disease. The model was developed using data from January 2000 to December 2012 then validated using disease notification data on reported cases for the following year. RESULTS: Monthly numbers of the mosquito Culex annulirostris (ß=0.033, p<0.001) and total rainfall (ß=0.263, p=0.002) were significant predictors of RRV transmission in the study region. The forecasted RRV incidence in the predictive model was generally consistent with the actual number of cases in the study area. CONCLUSIONS: A predictive model has been shown to be useful in forecasting the occurrence of RRV disease, with increased vector populations and rainfall being important factors associated with transmission. This approach may be useful in a public health context by providing early warning of vector-borne diseases in other settings.

Citizen science and smartphone e-entomology enables low-cost upscaling of mosquito surveillance.

Mosquito surveillance remains a cornerstone of pest and disease control operations globally but is strongly limited in scale by resources. The use of citizen science to upscale scientific data collection is commonplace, and mosquito surveillance programs have begun to make use of citizen scientists in several countries, particularly for exotic species detection. Here we report on a proof of concept trial in southern Australia for a citizen science mosquito surveillance program characterised by fixed point trapping with BG GAT devices and remote mosquito identification through emailed images, which we term 'e-entomology'. In a study with 126 participants, we detected mosquito seasonality with peak abundance in mid-summer (1.78 mosquitoes per trap per day), weather correlations (positive correlation with maximum temperature, r = 0.41) and a diversity of species (15 of 22 known species in the region) in a metropolitan setting. Whilst we demonstrated that the costs of a citizen science program is only about 20% of a comparable professional surveillance program, the mosquito community sampled by citizen scientists was biased towards container-inhabiting species, particularly Aedes notoscriptus. This is the first time fixed-point mosquito trapping has been combined with citizen science e-entomology to deliver comprehensive surveillance of urban mosquitoes.

Regional Comparison of Mosquito Bloodmeals in South Australia: Implications for Ross River Virus Ecology.

Ross River virus (RRV) is responsible for the most notifications of human arboviral infection in Australia. Seroprevalence and experimental infection studies have implicated macropods (e.g., kangaroos) as the major reservoir hosts. However, transmission ecology varies spatially, and infections in urban areas have prompted the question of what animals serve as reservoirs in regions where macropods are scarce. In South Australia (SA), human infection rates for RRV vary greatly by region as do vector and reservoir abundance. We hypothesized that mosquito abundance and feeding patterns would vary among ecoregions of SA and could help explain divergent human case rates. To test our hypothesis, we amplified and sequenced a 457 base pair region of the cytochrome B segment of mitochondrial DNA from blood fed mosquitoes collected in three main ecoregions of SA and identified sequences using a BLAST search in NCBI. Domestic livestock made up the vast majority of bloodmeals from the region with the highest human infection rate. Livestock are generally not considered to be important reservoir hosts for RRV, but our results suggest they may have a role in transmission ecology in some places. Surprisingly, none of the 199 bloodmeal samples were identified as macropod in origin. In the context of these findings, we consider the possible RRV vectors and reservoir hosts in these regions and propose that diverse spatial and temporal transmission ecologies occur in SA, depending on vector and reservoir availability.

Mosquito communities with trap height and urban-rural gradient in Adelaide, South Australia: implications for disease vector surveillance.

Understanding the factors influencing mosquito distribution is important for effective surveillance and control of nuisance and disease vector mosquitoes. The goal of this study was to determine how trap height and distance to the city center influenced the abundance and species of mosquitoes collected in Adelaide, South Australia. Mosquito communities were sampled at two heights (<2 m and ~10 m) along an urban-rural gradient. A total of 5,133 mosquitoes was identified over 176 trap nights. Aedes notoscriptus, Ae. vigilax, and Culex molestus were all more abundant in lower traps while Cx. quinquefasciatus (an ornithophilic species) was found to be more abundant in high traps. Distance to city center correlated strongly with the abundance of Ae. vigilax, Ae. camptorhynchus, Cx. globocoxitus, and Cx. molestus, all of which were most common at the sites farthest from the city and closest to the saltmarsh. Overall, the important disease vectors in South Australia (Ae. vigilax, Ae. camptorhynchus, Ae. notoscriptus, and Cx. annulirostris) were more abundant in low traps farthest from the city and closest to the saltmarsh. The current mosquito surveillance practice of setting traps within two meters of the ground is effective for sampling populations of the important disease vector species in South Australia.

Environmental and entomological factors determining Ross River virus activity in the River Murray Valley of South Australia.

OBJECTIVES: 1) To determine whether environmental and mosquito abundance variables could be used to explain fluctuations in the activity of Ross River (RR) virus, in the River Murray Valley of South Australia (SA). 2) To develop models at the local government spatial scale to understand local variability in RR activity factors. METHOD: Notification data of RR virus positive serology, mosquito surveillance, meteorological and river height data were analysed for the period 1999 to 2006. Stepwise multiple regression was used to determine significant environmental factors and to create descriptive models. RESULTS: The three models developed for different regions of the Valley explained significant amounts of variation in notification rates (R(2) 0.77 - 0.98). Regional variation in the models was observed, with differences in significant mosquito species evident. Rainfall was a significant predictor of RR virus activity in two of the models, while the height of the River Murray was significant in the third. An overall model for the entire SA section of the Valley contained only time-lagged mosquito abundance variables (R(2) 0.52). CONCLUSION: Although rainfall, river height and mosquito abundance are significant factors in determining RR virus activity, there are regional differences in this relationship. IMPLICATIONS: The regional variability of RR virus activity drivers has been defined, and has implications for the forecasting of future activity in this part of SA. The models provided here can provide the foundation for an effective RR virus early warning system, but only if criteria for action, lines of responsibility and the resources required have been determined.