Is it safe to swim at my favourite river?

Sharing is caring, but not particularly so, when you must share your favourite bathing spot with bugs that make you sick! Freshwater bodies are susceptible to contamination and can serve as a vehicle for transmission of microbial pathogens. The presence of these pathogens is usually “indicated” by the levels of faecal indicator bacteria (FIB, e.g. E. coli or enterococci) in the water, because the measurement of the pathogens themselves is difficult and costly. In New Zealand and many nations of the world, policy framework documents are in place to support monitoring programmes that warn the public when FIB levels at swimming spots exceed certain thresholds. These thresholds have been previously demonstrated to be indicative of heightened probabilities of the presence of potentially infective pathogens.

Perhaps the million dollar question any swimmer wants to ask is: “Is it safe to swim at my favourite river?” The answer to this question is tricky, because it depends on time scale and could be interpreted to mean any of the following:

• Is it safe to swim TODAY?
• Is it safe to swim NOW?
• Is it generally safe to swim (MOST OF THE TIME) at my favourite river?

To help the public answer these crucial questions, regional authorities place particular emphasis on swimming season monitoring of FIBs. Historically this monitoring was done only or twice once a month during summer. More recently, the recommended frequency of monitoring was revised to weekly, to capture a ‘recent snapshot in time’. Last year, a TV3 news item promoted the LAWA website as a place  ‘you can check safe swimming spots’ throughout New Zealand. Whilst the weekly monitoring during summer months is certainly an improvement over the status quo, the claim that  ‘even if LAWA’s data is a week or two out of date,… Kiwis should still consider the last update useful information’ does concern me, as there is the potential to underestimate risk. It is important to note that microbial populations fluctuate over much shorter timescales than a week in response to varying meteorological conditions and other factors (Figure 1).

Figure 1. A) Daily rainfall accumulation. B) River flows, turbidity and faecal indicator bacteria in the Sherry River. Source: Cornelisen et al 2011.

Even if the sampling frequency is increased, delays between sampling and reporting can hamper dissemination of timely information that is critical to preventing water-borne illnesses. The MfE mandates that Councils have communication strategies as part of their monitoring programmes to ensure the public is informed of a health risk at a beach or river as soon as possible when E..coli levels exceed single sample bathing water criteria. But then, how soon is soon?

Swimming advisories form part of such communication strategies.  In reality, however, swimming advisories only indicate that ‘it may have been safe/unsafe to swim in the past.’

By the time we have put up the warning signs about a river or beach with FIB above the guidelines (based on a sample collected 2 days prior), it has almost always dropped below guidelines [Trevor James, Tasman District Council, pers. comm. 2016].

The way forward

What then is the way forward?  Monitoring E coli continuously would be prohibitively expensive as well as being impractical. Emerging technologies such as environmental biosensors show significant potential for continuous E coli monitoring but the development and field validation required before such technology can be reliably deployed could take some time.

A more readily available strategy (until reliable continuous monitoring is in place), is to combine existing monitoring programmes with predictive models. This approach has been used for more than a decade in the USA but is new in New Zealand.  The reasons for this may be due in part to the misconception that you need to accurately predict individual E. coli values rather than whether the E coli values will exceed a threshold (Figure 2).  Predictive model performance is better assessed by a combination of specificity (the proportion of true non-exceedances of bathing water standards),  sensitivity (the proportion of true exceedances of bathing water standards) and accuracy (the proportion of true exceedances and non-exceedances correctly predicted by the model).

Figure 2: This simple regression model of observed vs modelled E. coli has a high R squared value but only correctly predicts four out of the eleven exceedances of the bathing water standard. Source: C. Dada.

An advantage of predictive models based on specificity. sensitivity, and accuracy is that they can provide ‘nowcasts’ and forecasts of FIB concentrations, based on current/ future land use and meteorological scenarios. Another advantage is that they use easily measured environmental variables such as rainfall, wind speed, temperature, turbidity, and water clarity, as predictors. The models estimate densities of FIB but, more importantly, also estimate the probability that the single sample bathing water criteria will be exceeded. Such models have been applied to a number of recreational bathing sites in New Zealand (e.g., Lake Rotorua). Another major advantage of the FIB predictive models is that they can be both calibrated and validated using existing (but independent) historical data. They are site specific because both the hydro-meteorological conditions and sources of E. coli differ between sites. However, they are quick to develop, can be progressively tested and upgraded with each season’s data, and can be readily incorporated within a real-time system that potential swimmers can check before they decide to take the plunge!

Going back to the question every potential swimmer seeks an answer for. Is it safe to swim at my favourite river? By including predictive models to complement monitoring efforts, we can satisfactorily answer that question to meet the requirements of Councils, heath officials and the general public.

Christopher Dada

Reference: