A mixed-method examination of risk factors in the truck-to-cistern drinking water system on the Beardy’s and Okemasis First Nation Reserve, Saskatchewan

Abstract

The provision of safe drinking water is a key driver of public health and a pressing health issue facing First Nations communities in Canada. Contaminated water is a perennial issue for reserve communities across the country despite numerous government investments. Many First Nations communities rely heavily on cistern use for their drinking water supply; however, bacterial contamination within these systems is frequent and a common reason for household boil water advisories. The sources of contamination during the process of hauling water to cisterns in First Nations communities have received limited attention in academic research. The purpose of this research is to identify the risks to water quality through the truck-to-cistern water system. In partnership with a Saskatchewan First Nations community, drinking water quality was monitored in the treatment plant, in delivery trucks and at 142 household cisterns and taps from July to October, 2014. Risks to water supply were identified through monthly water sampling and laboratory analysis, key informant interviews, and observation. Coliform contamination in trucks, cisterns and taps was most common during August. Total coliforms were more likely to be found in cisterns compared to household taps and samples from trucks. Chlorine residuals were lower in household tap samples than in cisterns for August and September. Together with the community, investigators identified contamination and bacteriological growth in cisterns and household distribution systems, and variable levels of total chlorine concentrations depending on month and site of sampling. Recommendations are provided for advancing guidelines on management of truck-to-cistern drinking water supply chains in First Nations.

L’approvisionnement d’eau potable est un facteur de la santé publique et un enjeu de santé pressant pour les collectivités des Premières nations au Canada. Malgré de nombreuses évaluations gouvernementales, l’eau contaminée est à maintes reprises considérée comme un problème permanent pour de nombreuses collectivités à travers le pays. Les collectivités des Premières nations dépendent fortement de l’utilisation de la citerne pour l’approvisionnement d’eau potable, mais la contamination bactérienne dans ces systèmes est courante et un message important pour émettre des avis sur l’eau bouillante dans les collectivités. Le transport de l’eau vers les citernes en tant que partie d’une chaîne d’approvisionnement d’eau potable est connu sous le nom de « problème des prairies » et a néanmoins reçu une attention limitée en termes d’initiatives gouvernementales et de recherche universitaire. L’objectif global de cette recherche est d’identifier les risques potentiels pour la qualité de l’eau à travers la chaîne d’approvisionnement de l’eau transportée par camion au stockage de la citerne. En partenariat avec la collectivité, la détérioration de la qualité de l’eau potable et les risques pour l’approvisionnement en eau de camions et citernes ont été analysés dans 142 ménages d’une Première nation de la Saskatchewan en 2014. Les taux de contamination dans les camions et les citernes étaient les plus élevés pendant les mois d’été, les plus hauts dans les citernes et les robinets, et le bas niveau des résidus de chine ont été trouvés dans les échantillons de prise de ménage. Les risques pour l’approvisionnement en eau grâce à l’échantillonnage de l’eau et à l’analyse de multiples paramètres de l’eau, les entretiens avec les informateurs et l’observation ont été identifiés et incluaient une turbidité élevée dans le transport par camion, la contamination et la croissance bactériologique dans les citernes et les systèmes de distribution des ménages, la mois échantillonné et lieu d’échantillonnage. Des recommandations pour l’avancement des lignes directrices sur la gestion et le suivi des chaînes d’approvisionnement en eau potable des citadins des Premières nations sont discutées.

Keywords:

• water quality
• truck-to-cistern
• First Nations

Introduction

Access to safe drinking water is vital for human health and is a major public health and human rights issue in countries across the globe (Cook and Bakker 2012; Bain, Wright et al. 2014). Strategies to improve safe drinking water supplies have been implemented over several decades. One example is the September 2000 United Nations Millennium Development Goals which have been adopted to improve access to piped water and decrease disparity between urban and rural dwellers (Bain et al. 2012). In Canada, the multi-barrier approach to protect drinking water by provincial regulators emerged after the 2002 Walkerton water crisis (Canadian Council of Ministers of the Environment [CCME] 2002, 2004). As a result, improvements to centralized piped distributed water supply systems have been made, yet drinking water safety concerns remain for those on decentralized non-piped systems (e.g. wells, and truck-to-cistern) (Prudham 2004). Decentralized non-piped distributed water systems provide drinking water to Indigenous reserve communities in Canada where centralized piped distribution systems or private wells are not economically or technically feasible, as determined by Indigenous and Northern Affairs Canada.

As a result of rights-based strategies, Canada has implemented a multi-agency system in which local, regional, provincial and federal levels of government work together to establish drinking water quality guidelines; source water protection plans; certification systems for infrastructure, operations and maintenance; innovative structures for processing and delivery; and educational curricula on drinking water safety. Responsibilities are shared among individual community members up to federal government levels to work toward the common goal of safe drinking water (World Health Organization 2004; Reid et al. 2014). Opportunism does emerge, however, due to supply and demand characteristics for water in rural areas and reserves.

Rural dwellers including those living on First Nations reserves often rely on private wells, truck-to-cistern systems or a combination of both, and on bottled water delivery in extreme cases. In the Canadian Prairie provinces, it is estimated that one fifth of households on reserves used truck-to-cistern systems (Duncan and Bowden 2009). While reserves may have water treatment facilities, the transport of the water to storage containers near point of use and the maintenance of the storage and distribution systems (e.g. cisterns, pumps, household pipes) depend on contractors or subcontractors who may be trained or untrained and work at varying costs. While this opportunism is valuable for (1) managing short-term consumer needs, (2) increasing access to water where none would be easily available, (3) incurring less expense for in-house manufacturing or service provision, and (4) improving efficiency, it also has drawbacks. These drawbacks include having minimal information exchange during water delivery, little care given to ensuring safety or quality standards, few built-in redundancies in case of emergency or supply failure, and no aligned goal(s) for all members of the delivery chain (Vandenbosch and Sapp 2010). For the provision of safe drinking water, the use of opportunistic vendors for cistern-based systems could increase risks to human health (Wagner and Bode 2008; Lindovsky and Krocova 2015). Health risk is defined here as being an increase in any attribute, characteristic or exposure of an individual that intensifies the chance of developing a disease (WHO 2018).

In this paper, the risks associated with the truck-to-cistern water supply chain are explored in one Saskatchewan First Nations community. The results contribute to the concerns brought forward in four recent works on disparities in the provision of safe drinking water for Indigenous communities in Canada. Baird, Summers, and Plummer (2013) described the nature and extent of water contamination risks in one Canadian reserve community using cistern systems, but called for more case studies to affirm their findings. Daley et al. (2014) discussed how northern Indigenous communities face challenges of monitoring water quality, record-keeping and data management, maintaining chlorine levels to the taps, and shortages of certified operators, similar to challenges of rural communities elsewhere, but also have contextual factors stemming from local culture and value systems, and from the northern climate. Farenhorst et al. (2017) demonstrated that proteobacteria prevalence changed in truck-to-cistern water supplies depending on sample sites, water treatment plants, truck hauling tanks, buckets, wells, cisterns, piped systems and taps in homes on a reserve in Manitoba. Finally, Plummer et al. (2013) highlighted the regulatory gaps that exist for non-piped water systems on reserve in Canada and associated risks to public health.

The case study shared herein describes the truck-to-cistern water supply system in one community. Regulatory gaps across the supply chain are highlighted which have the potential to increase risks to consumers and on-reserve governments. Potential enhancements to the regulations and the supply chain are discussed which could provide a pathway to alleviating some of the public health and household risks associated with the current approach. The paper begins with background information on cistern-based water systems and then describes the current study and results. A discussion of the findings with a focus on current regulations and gaps follows.

Literature review: An overview of risks and regulations in cistern-based water systems in Canada

Cisterns: characterization and prevalence

Cisterns are used extensively in Canada, on and off reserve, for private and public drinking water provision (Figure 1). A cistern is defined as a container used to store household potable water, intended for long-term storage for weeks to months, representing the final point at which potable water reaches the consumer (Agriculture AgriFood Canada [AAFC] 2006; Mohamed & Gad 2010; Baird, Summers, and Plummer 2013). Cisterns are growing in popularity in urban centers for rainwater collection, and may serve to ease drought pressures in areas under threat from climate change (Campisano et al., 2017), but in the rural and reserve sector, cisterns are used out of necessity (Baird, Summers, and Plummer 2013; Farenhorst et al. 2017). Rural dwellers are required to haul their own water, or pay to have their cistern water delivered from other private sources, municipally controlled and provincially managed treatment plants. First Nations reserve dwellers are dependent on their chiefs and council members for the delivery, and these councils face disparities in finding service providers to do the hauling, or local people to train, retain and serve the reserve.

Figure 1. Truck-to-cistern drinking water supply chain.

An assessment by Neegan Burnside Ltd (2011b) found that in Canada, 13.5% of reserve houses depend on the truck-to-cistern drinking water supply chain. An estimated 21% of the houses on reserve in Saskatchewan are dependent on cisterns (Neegan Burnside Ltd. 2011a). Duncan and Bowden (2009) describe hauling water as the ‘Prairie Problem’, and yet cisterns on First Nations have received limited attention in terms of government initiatives and academic research (Baird, Summers, and Plummer 2013). Contamination in cisterns is common and is one of the leading causes of boil water advisories in First Nations communities (Duncan and Bowden 2009; Moffatt and Struck 2011; Baird, Summers, and Plummer 2013; Farenhorst et al. 2017). Questions persist regarding who has the responsibility to maintain and manage these water systems and whether current testing protocols are adequate (Saskatchewan Research Council and University of Saskatchewan [SRC and USASK] 2013; Fernando et al. 2016).

Cisterns: risks and regulations

In developed countries, the presence of a water distribution system is perceived as a sign of improved water quality, although it does not imply that the water is free of risks and is adequate for human consumption (Schafer 2010). Small drinking water distribution systems supplying smaller communities (< 5000 people) and private systems in Canada have a higher occurrence of water-borne disease outbreaks than do large systems (Moffatt and Struck 2011; Baird, Summers, and Plummer 2013). Oftentimes, water leaving treatment systems or arriving at a residential cistern is microbiologically safe; however, contaminants may enter a distribution system after treatment, during travel, or in household storage (Farenhorst et al. 2017). There is the potential for localized contamination through many pathways. Point contamination can affect drinking water through inadequate initial treatment, during the transportation process (AAFC 2006), in the course of transfer to storage containers, with poorly maintained infrastructure (Schafer 2010; Farenhorst et al. 2017), and by existing contaminants in the household distribution system (Health Canada 2005; Duncan and Bowden 2009; Schafer 2010; Baird, Summers, and Plummer 2013). Cisterns present an additional challenge due to microbial growth occurring in the cistern tank itself over time (Baird, Summers, and Plummer 2013; Farenhorst et al. 2017). Retaining large volumes of water at the household level for long periods of time results in the growth of contaminant microorganisms, inadequate disinfection residuals, and increased temperatures (Schafer 2010; Ohkouchi et al. 2013; Farenhorst et al. 2017). When drawdown in the tank reduces the volume of water, or delays in recharging tanks or disinfecting tanks occur, risk levels increase (Fortier 2010). These factors can all contribute to water parameters that are unacceptable for human health (Health Canada 2012a). Boil water advisories resulting from contaminated cisterns have ranged from short-term (weeks to months) to years and have not decreased in frequency or duration in decades (SRC and USASK 2013; Farenhorst et al. 2017).

To reduce risks, monitoring via regular sampling is necessary from the water source to the consumers’ tap. Incomplete coverage of water quality testing across Canada hampers monitoring effectiveness (Dunn et al. 2014). The Canadian Drinking Water Quality Guidelines have been adopted in full in only one provincial/territorial jurisdiction in Canada; in majority in two jurisdictions; and in part in five out of 13 jurisdictions, not including Indigenous self-governed areas (Dunn et al. 2014). Health Canada’s (2012b) Guidance for Designing, Installing, Maintaining and Decommissioning Drinking Water Cisterns in First Nations Communities South of 60° is the most often cited policy document for cistern systems. Provinces and territories can adopt the Health Canada guidelines, and/or develop their own. Baird et al. (2013) provide a comprehensive list of federal, provincial and territorial legislation for cisterns and water transport; however, they demonstrate that the legislation varies in content, schedule of testing, and specificity of recommended water quality tests at local, regional and provincial levels.

No binding federal regulations exist for cisterns on reserves. The Health Canada (2012c) document provides haulers with information on how to safely collect, transport and distribute water, and lists criteria and practices for managing the truck-to-cistern systems such as the expected roles and responsibilities of chiefs and council members, drinking water monitors and health officers, water haulers and home occupants. In addition, the Protocol for Safe Drinking Water in First Nations Communities (hereafter, the Protocol) sets out a regulatory framework that considers trucked water systems as an extension of piped systems (Indigenous and Northern Affairs Canada 2010). The Protocol also specifies that these systems should meet the water quality requirements of a piped system, use tanks that are fabricated from suitable materials to transport water, and be hauled by an operator with adequate training (INAC 2010). As others have noted, the Protocol complicates understanding of responsibilities across different jurisdictions, and sets out guidelines that are not informative of risk management, or easily enforced (Baird, Summers, and Plummer 2013; Farenhorst et al. 2017). According to the Protocol, for example, trucked water systems are only required to report on one chlorine residual sample per water delivery day, yet could make multiple deliveries in a day, each time increasing the risk of exposure to multiple sites where contamination is possible. In addition, no records of truck hauler certifications are catalogued at the reserve or federal level in Canada.

Reserve governments have the authority to manage natural resources including water, according to any existing local by-laws, which vary from community to community. The ad hoc regulatory situation on reserves in Canada differs from that of other developed countries where comprehensive source-to-tap regulations exist at the federal level for cistern use, and in some cases are built on and made more stringent by state or local governments (e.g. Australia, United States Environmental Protection Agency [US EPA], Iceland, and Ireland) (Hrudey 2011; Baird, Summers, and Plummer 2013). In other regions of the world such as Saudi Arabia and Spain, individual citizens are responsible for the maintenance and water quality of their private cisterns, and some have demonstrated innovations for ensuring high-quality cistern-based drinking water (Guizani 2016; Vallès-Casas et al. 2016).

These examples provide lessons for developing a system of regulations and practices in Canada for cisterns on reserve. First, when there is an established drinking water safety plan, and the chain of responsibility is clear, there is a reduction in the frequency of drinking water advisories in small, private citizen-maintained systems (Gunnarsdóttir 2012). Second, participatory, co-development and social regulations approaches for small drinking water management programs in rural communities have enhanced health outcomes in comparison to top-down and centralized approaches (Sharma 2009; Daniell 2012; Reddy et al. 2014). Third, specific monitoring results, examination of current knowledge of health outcomes and risks, and user practices in truck-to-cistern systems are needed to support the development of locally relevant guidelines and regulations (Lye 2002; Bain, Cronk et al. 2014; Constantine et al. 2017).

Cisterns: Current supply chain treatment practices

Several practices exist and contribute to drinking water safety in truck-to-cistern systems in Canada. Ensuring high-quality source water is the first step for providing potable water on reserve. Source water protection planning on reserves is difficult and requires local commitment and engagement, adequate resources and funding to support the process, local leadership and capacity to complete the planning process, and long-term commitment to enforcement of the plan (Phare 2009; Basdeo and Bharadwaj 2013). While source water protection plans are mandatory in some provinces and include regulatory and non-regulatory tools, these plans may not be a priority in First Nations communities due to more immediate issues. A lack of resources and poor relationships with surrounding municipalities may also contribute to prioritization of source water planning. In addition, even when binding federal legislation regarding the protection of source water on reserves exists, little support and few resources for complying are available to reserve communities, and watershed boundaries often extend beyond those of the reserve itself (de Loë et al. 2010).

Water is typically treated in plants on reserve or at nearby municipal systems before being trucked to cisterns. Water quality from the plant is dependent on many factors, including availability of replacement parts, current technology, capacity of the plant to meet the community’s needs, and whether the regulations and monitoring plans for the treatment plants are followed (Baird, Summers, and Plummer 2013). Drinking water safety on reserves using cistern systems relies heavily on protective actions, and on disinfection practices during transport and transfer to cisterns for storage and distribution (Mohamed and Gad 2010). The most commonly used disinfectants in water distribution systems are chlorine and chlorine compounds (Carrico and Singer 2009; Mohamed and Gad 2010; Ramos et al. 2010; Schafer 2010; Sun et al. 2013). Chlorine added to water in its molecular or hypochlorite form initially undergoes hydrolysis to form free chlorine (consisting of aqueous molecular chlorine, hypochlorous acid (HOCl), and hypochlorite ions (OCl⁻). Free chlorine is the chlorine available for disinfection and combined chlorine is chlorine that has reacted with nitrogen-containing compounds to form chloramines (American Public Health Association et al. 2005; Schafer 2010). The chlorine residual is the amount remaining after oxidation due to reaction with organic matter and metals in the water. If the dosing rate of chlorine is below guidelines, there may be insufficient residual at the faucet, resulting in bacterial regrowth if the water is contaminated in the storage or distribution system (Ramos et al. 2010; Ohkouchi et al. 2013; Sun et al. 2013). The chlorine residual within the water supply depends on numerous factors including reactivity with water; pipe wall material; hydraulic flow regimes; duration of residence; control of pH; temperature; and the presence of organic matter, microorganisms and biofilms and corrosion of pipe walls (Ramos et al. 2010; Coleman et al. 2013; Ohkouchi et al. 2013). A lack of chlorine residual often indicates that contaminants are present.

The Guidelines for Canadian Drinking Water Quality suggest free chlorine concentrations should range from 0.04 to 2.0 mg/L (Health Canada 2009). Guidelines for free chlorine residuals vary by province, territory and jurisdiction in Canada and are determined by the authority responsible. In other countries, strict guidelines are enforced; Japan, for example, legally requires a chlorine residual of 0.1 mg/L to be maintained out of each faucet, but their waterworks managers maintain a chlorine residual of around 0.3–0.4 mg/L at each faucet (JWWA 2010 in Ohkouchi et al. 2013). In general, a free chlorine residual of 0.2 mg/L is considered a minimum level for the control of bacterial regrowth in the distribution system (Health Canada 2012a).

The transportation and transfer process is also currently unregulated on reserve apart from the Health Canada guidelines (2012). In addition, Indigenous and Northern Affairs Canada (INAC) produced Design Guidelines for First Nations Water Works as a general guide for engineers in the preparations of plans for the design and construction of water treatment facilities on reserves (2006). Both sets of guidelines and the Protocol were created to reduce potential public health risks associated with trucked water delivery and cistern storage systems; however, problems with these systems remain more frequent than for piped distributions (Moore 1999; Levangie 2008, 2009; McCullough and Farahbakhsh 2012). Thompson (2016) reveals that small, non-piped systems in Saskatchewan are more likely to have drinking water advisories that stem from operational causes, regardless of whether or not the operator was fully certified.

Social and political deficiencies in a truck-to-cistern system were identified in a Saskatchewan First Nation, that compromised the management of the water system (Lebel and Reed 2010). In addition, differences in cultural and economic approaches to water in Indigenous communities in Canada influence the priorities of community decision makers regarding risk management and intervention development (Daley et al. 2015). The authors add that these approaches are better understood when both quantitative and qualitative data are sought in studies of small, community-based drinking water systems (Daley et al. 2015).

A final set of practices identified as contributing to drinking water safety in reserve truck-to-cistern systems involves consumer preferences, bottled water consumption and home treatment. An estimated 48.6% of Canadians in a cross-sectional meta-analysis of seven community-level studies used an in-home treatment method (e.g. jug filter, tap filter, water softener, boiling, reverse osmosis system, in-line filter, fridge filter, ultraviolet disinfection, iron removal system, ozone disinfection system, and/or candle filter) as a final step in the drinking water supply chain (Roche et al. 2012). Studies of consumer preferences have shown that taste (particularly organoleptics), health risk concerns, regulations, trust, perceived control, cultural background, past experiences and location are implicated in the consumers’ choice to use tap water for drinking water (Doria 2006; Dupont et al. 2014). Researchers suggest that increased education with regard to the spectrum of taste, odour and appearance of safe drinking water is needed to help consumers overcome negative risk perceptions (Owen et al. 1999; Johnson 2003; Doria 2010).

The production of safe drinking water for reserve cistern systems requires many sequential steps, each following diverse sets of non-binding guidelines and conventional practices, to deliver water from source to storage/cistern and then to tap while maintaining water quality. Few studies have measured water quality indicators along each step in the supply chain for cistern systems on reserves in Canada as a means to evaluate implementation of existing guidelines and how effective they are in practice. To add to recent work identifying growth of protobacteria in truck-to-cistern systems in Manitoba (Farenhorst et al. 2017), this study examines the process of water delivery, factors affecting chlorine residuals, and sites of potential contamination in a truck-to-cistern-dependent reserve in Saskatchewan.

Methods

Study site

Beardy’s and Okemasis First Nation (BOFN) is a reserve in Treaty 6 within rural north central Saskatchewan, 85 km north of the city of Saskatoon. Approximately 1100 people live on the reserve, of the 2900 registered members of BOFN (SRC and USASK 2013). BOFN is mid-sized compared with other reserves in Saskatchewan (population range: 17 to 5482 people; Neegan Burnside Ltd 2011a). Drinking water systems on First Nations are under federal jurisdictions and do not require certificates or licenses of classification; however, they may be notionally classified. Notional classifications are not formal classifications, but they identify the type of water facility to which the federal ministry deems them comparable across other jurisdictions (i.e. provincially, or municipally). The classification of treatment facilities is based on a range of points determined by the degree of difficulty in operating the facility. A Level 1 facility is generally sourced from groundwater, serves a population of less than 500 and may include treatment options such as iron and manganese removal facilities, water softening facilities using ion exchange, ultraviolet treatment () for virus inactivation, or filtration/ chlorination facilities, and employ a Level 1 certified water treatment operator. The BOFN community’s potable water, derived from three groundwater wells, is treated at a Level 1 water treatment facility (established in 1992) and operated by a primary certified operator. Greensand filtration followed by chlorine disinfection are the processes used in the facility.

There are 142 homes on a piped distribution system from the treatment facility, 142 homes with cisterns filled by truck-to-cistern with water from the plant, and 53 homes on private wells. Based on historical data, the quality of the treated water from the facility has been within acceptable levels for many years. Despite having an effective water treatment facility, 18 cisterns experienced bacterial contamination between July 2012 and June 2013, and two further cisterns have been on boil water advisories since 2009 (SRC and USASK 2013). Four private wells are filled with water trucked from the treatment plant for household use. In years before this study, a community-based project examined the water quality in the community and found that there was inadequate and irregular disinfection and cistern cleaning, and many cisterns were in need of maintenance (SRC and USASK 2013). A further 15 cisterns had cracked or improper seals, and contained debris (dirt, deceased animals). The study also found that five private wells had owner-funded and maintained in-house treatment systems, while houses served by cisterns did not. The report stated that 33% of surveyed residences indicated that their cistern had never been cleaned, and 43% indicated that their cistern had been cleaned once in the last 5 years (SRC and USASK 2013). The subsequent projects described below arose from the SRC and USASK (2013) findings and were supported by the chief and council of BOFN.

Qualitative interviewing

Between April 2012 and August 2014, and as part of a larger research program examining the perspectives of members of nine Saskatchewan reserves on the health risks associated with their on-reserve drinking water treatment systems, 23 English-language interviews were conducted at the BOFN reserve by a graduate student. Participants included eight men and 15 women who, at the time, lived and worked in the community, had lived in their house on average for 19.3 years (range 4–35 years), with an average household of 4.2 people (range 1–7). Interviews were thematically coded by two independent coders following the Boyatzis (1998) protocol. Codes were compared with the subsequent laboratory test results from the water quality sampling described below to gain a holistic understanding of threats in the water distribution system. The interview guide was co-developed and approved by the community chief and council and by the University of Saskatchewan Ethics Board prior to data collection (BEH 12-121). On sharing of the initial data from the earlier interviews, the community decided to explore their water quality as described next.

Water quality sampling

An initial observational assessment of all cisterns was completed in June 2014 to categorize BOFN’s cisterns as high or low risk depending on the criteria developed by Health Canada and by locally held records for water quality, chlorine decay rates, and re-contamination rates (Health Canada First Nations and Inuit Health Branch 2012). Each house was numbered; then a representative sample (n = 60 cisterns of 142, half high risk and half low risk) was selected using an online random number generator. Water samples were collected and analyzed between June and October 2014 in the truck-to-cistern drinking water supply system at up to four locations in addition to the regularly sampled water treatment plant: (1) the site of transfer from the water treatment plant into the truck at a water treatment facility (before driving); (2) water transported in water holding trucks at the cistern site (after driving); (3) water inside a residence cistern sampled from within the cistern; and (4) water collected at the consumers’ tap after filter removal, nozzle disinfection and 3 min of running (Tables 1 and 2). Project personnel observed that trucks were used opportunistically to deliver water on reserve during the sampling period. It was not possible to sample every truck used over the 4-month period as not all possible trucks were used on each sample day. However, samples were collected from all accessible trucks on the sample days. Water test parameters included microorganisms, turbidity and chlorine. Eight samples were collected from the most distant outlet within the plant to align with the guidelines for microbial testing in the Protocol for Centralized Drinking Water Systems in First Nations communities (INAC 2010).

Table 1. Sampling and handling requirements revised from Standard Methods for the Examination of Water & Wastewater 2005.

Parameter	Suitablecontainer	Minimum sample size	Sample type	Preservation	Sample-holding times
					Recommended	Regulatory
General chemical/water quality – on-site analysis Equipment – YSI multi-parameter probe, portable colorimeter
Chlorine: total, free, residual	Plastic, glass	500 mL	Grab	Analyze immediately	0.25 h	0.25 h
pH	Plastic, glass	50 mL	Grab	Analyze immediately	0.25 h	0.25 h
Temperature	Plastic, glass	-	Grab	Analyze immediately	0.25 h	0.25 h
Turbidity	Plastic, glass	100 mL	Grab	Analyze same day: store in dark up to 24 h, refrigerate	24 h	48 h
Bacteriological parameters – Saskatchewan Disease Control Laboratory, Environmental Services, Regina, SK
Total coliform	*Sterile plastic	250 mL	Grab	Refrigerated conditions	48 h	48 h
E. coli	*Sterile plastic	250 mL	Grab	Refrigerated conditions	48 h	48 h

^* The 250-mL bacteriological bottles are sterilized and are obtained directly from the Saskatchewan Disease Control Laboratory (5 Research Drive, Regina, SK, Canada, S4S 0A4).

Grab samples are single samples collected at a specific spot, over a short period of time (seconds or minutes).

Sample-holding time is the time period from sample collection to sample analysis.

Table 2. Water sampling timeline – all water quality parameters.

Water sampling timeline^*	Sampling site
2014 sampling period	(i) Source: water treatment facility	(ii) Water holding truck delivery	(iii) Cistern	(iv) Consumers’ taps, cold	Sample totals
June	2	2 trucks × 10 days = 20	60	60	142
July	2	2 trucks × 10 days = 20	60	60	142
August	2	2 trucks × 10 days = 20	60	60	142
September (or October)	2	2 trucks × 10 days = 20	60	60	142
Total samples:	8	80	240	240	568

^* Timeline and sampling analysis were confirmed following consultation with Saskatchewan Disease Control Laboratory Supervisor (5 Research Drive, Regina, SK, Canada, S4S 0A4).

The Standard Methods for the Examination of Water and Wastewater were used to guide sampling and analysis (Water Environmental Federation & American Public Health Association 2005). Detection limits were assessed under the Protocol, and the Canadian Guidelines for Monitoring Public Drinking Water Supplies (Health Canada 2009). Samples were stored under refrigerated conditions in portable coolers and shipped within 24 h of sampling to the laboratory (as per Ohkouchi et al. 2013) at the accredited Saskatchewan Disease Control Laboratory (SDCL) in Regina. All tests were performed by certified employees, using Canadian Association for Laboratory Accreditation protocols in accordance with International Organization for Standardization (ISO) 17025. Detection limits for each of the tests fall within the guidelines set for Canadian drinking water. All sample containers and preservatives were provided by the SDCL. The laboratory also provided a detailed protocol for sample collection, pre-printed sample labels on each container, and chain-of-custody forms.

Water quality sampling was conducted weekly or monthly depending on the location and parameter (Table 2). Methods from the Standard Methods for the Examination of Water and Wastewater (2005), and tests accredited through the Canadian Association for Laboratory Accreditation (CALA) in accordance with ISO 17025, were used. Biological parameters of total coliform (membrane filtration, MF) and E. coli (MF) were analyzed at the Saskatchewan Disease Control Laboratory, in Regina, Saskatchewan. For quality assurance and control, the SDCL applied the most probable number (MPN) method to a random number of samples (Edberg et al. 1988; Cowburn et al. 1994). Samples were placed aseptically in sterile bacteriological bottles containing sodium thiosulfate obtained from the SDCL. Samples were stored under refrigerated conditions in portable coolers (at 5 ± 3 °C), kept in the dark, and shipped within 24 h after sampling to the laboratory (Ohkouchi et al. 2013) to minimize changes in populations and concentrations (Dutka and El-Shaarawi 1980). In the MF method a minimum volume (10 mL of the sample or dilution of the sample) is introduced aseptically into a sterile filtration assembly containing a sterile membrane filter (nominal pore size 0.2 or 0.45 μm). A vacuum is applied and the sample is drawn through the filter. All indicator organisms are retained on or within the filter, which is then transferred to a suitable selective culture medium in a Petri dish. Following a period of resuscitation, during which the bacteria become acclimatized to the new conditions, the Petri dish is transferred to an incubator at the appropriate selective temperature where it is incubated for a suitable time to allow the replication of the indicator organisms. Visually identifiable colonies are formed and counted, and the results are expressed in numbers of ‘colony-forming units’ (CFU) per 100 mL of original sample. The MPN method is based on an indirect assessment of microbial density in the water sample by reference to statistical tables to determine the most probable number of microorganisms present in the original sample. Each of the separate volumes drawn from the original sample is mixed with culture medium and incubated. The concentration of microorganisms in the original sample is estimated from the pattern of positive results (the number of tubes showing growth in each volume series). The MPN method was used for analysis of total coliforms and of E. coli.

The chlorine residual (Cl₂), total chlorine (mg/L), free chlorine (HOCl mg/L), and turbidity were measured in the field using a LaMotte portable DPD colorimeter (Health Canada 2009; Mohamed and Gad 2010). Water temperature (°C) and pH were also measured in the field, using YSI multi-parameter probes.

Statistical analyses

Generalized linear mixed models were used to look at the association between risk factors and outcomes of interest. Repeated measures on individual households and trucks were accounted for with random intercepts for house and truck as appropriate in all initial unadjusted, univariate and final multivariable models.

In the first model, factors associated with whether or not total coliforms were cultured were examined for samples from truckloads of water using a logit link function and assuming a binomial distribution. Risk factors considered in this model were sample month (July, August, September and October), water temperature, total chlorine, free chlorine and turbidity. A second, similar model was constructed to determine whether or not total coliforms were cultured from the cistern or tap from a sample of community households. Initially all factors were examined in bivariate or unconditional models for the presence of coliforms. Risk factors in which p < 0.20 were considered in the creation of the final multivariable model. The strength of association for significant risk factors was summarized as odds ratios (OR) with 95% confidence intervals (95% CI). Similar models were not constructed for E. coli results due to the very low frequency of isolation.

In a second series of models, risk factors of interest were evaluated for characteristics of both truck and household water (cistern and tap) including total chlorine, free chlorine and turbidity. These models assumed a normal distribution.

The differences between cistern and tap water were included for all household models. All final multivariable models included month of collection and any other significant risk factors (p < 0.05). Models were built in a stepwise backward direction, and examined for extreme or influential outliers, and for continuous outcomes the residuals were examined for normal distribution and homogeneous variance. The linearity assumption was also examined for continuous risk factors.

Results: Water sampling

Sampling and culture results

Eight water samples were collected from the farthest outlet of the water treatment plant during July to October 2014 on an established bi-weekly schedule. In addition, 82 samples were collected from 17 of the delivery trucks after they had been filled, but prior to leaving the treatment plant. The median number of samples per truck was three (minimum = 1, maximum = 14). The objective of truck sampling was to sample all of the trucks servicing households on the day of sampling. The random effects model accounts for variation among trucks due to confounding and clustering. In addition, 235 samples were collected and submitted from 61 cisterns, and 236 samples were collected and submitted from 61 matched household taps, for the same period. The median number of samples per cistern and household taps was four (minimum = 1, maximum = 5).

The proportion of samples in which total coliforms were identified were higher in cisterns (38.8%) and household taps (30.7%) than in samples collected from trucks (25.9%) and the plant (0%) (Table 3). Total coliforms were detected at least once in 91.7% of sampled trucks, 83.8% of cisterns and 75.3% of household taps sampled. The one truck where total coliforms were not detected at least once was only sampled twice during the study.

Table 3. Descriptive summary of water culture statistics, July–October 2014.

		Frequency
Variable	N	Positive samples	%	5th percentile	Median	95th percentile	Maximum
WATER PLANT
Total coliforms	8	0*	0.0	0.0	0.0	0	0
E. coli	8	0*	0.0	0.0	0.0	0	0
TRUCKS
Total coliforms	81	21	25.9	0.0	0.0	17	> 100
E. coli	81	0	0.0	0.0	0.0	0	< 1
CISTERNS
Total coliforms	232	90	38.8	0.0	0.0	44	> 100
E. coli	232	3	1.3	0.0	0.0	0	1
TAPS
Total coliforms	228	70	30.7	0.0	0.0	32	> 100
E. coli	228	2	0.9	0.0	0.0	0	9

^* No detectable coliforms or E. coli.

Using routine culture, E. coli was not identified in any of the truck samples, but was identified in three samples from one cistern and two samples from two household taps. Based on MPN culture of a subset of samples (Oblinger and Koburger 1975), a fourth cistern and second tap sample from the household were identified as positive for E. coli.

Total chlorine and free chlorine concentrations were slightly higher in samples collected from the trucks, and turbidity measurements were slightly lower than in household samples collected directly from the cisterns or taps (Table 4). Temperature variation was not associated with changes in chlorine concentrations.

Table 4. Descriptive statistics for water quality parameters, July–October 2014.

Variable	Average	Standard deviation	N	5th percentile	Median	95th percentile
TRUCKS
Total chlorine (mg/L)	1.41	0.23	82	1.12	1.41	1.81
Free chlorine (mg/L)	1.29	0.25	82	0.79	1.33	1.68
Turbidity (NTU)	0.11	0.13	82	0.01	0.07	0.36
Temperature (°C)	9.2	1.2	81	7.6	9	11.4
CISTERNS
Total chlorine (mg/L)	0.92	0.33	235	0.32	0.94	1.39
Free chlorine (mg/L)	0.81	0.32	235	0.17	0.84	1.29
Turbidity (NTU)	0.46	1.64	235	0.03	0.13	1.39
Temperature (°C)	10.8	1.4	235	8.8	10.7	13.2
TAPS
Total chlorine (mg/L)	0.92	0.73	231	0.19	0.91	1.39
Free chlorine (mg/L)	0.77	0.35	231	0.09	0.81	1.25
Turbidity (NTU)	0.71	3.45	231	0.02	0.12	2.14
Temperature (°C)	11.6	1.6	231	9.1	11.6	14.6

Factors associated with the presence of total coliforms

Based on the results of unconditional or bivariate analysis (Table 5), the variables considered in building the final multivariable model for factors associated with identifying total coliforms in truck samples included sample month, water temperature, and total and free chlorine concentrations. In the final multivariable model for the truck samples, total coliforms were more likely to be identified in the samples collected in August than the samples collected in October (OR 7.18) and in samples that had higher total chlorine concentrations (OR 1.68 per 0.1 mg/L increase) (Table 6).

Table 5. Unadjusted associations between potential risk factors and presence of total coliforms.

Risk factor	Odds ratio	95% CI		p
		Lower	Upper
Location: hruck
Sample month:				0.03
July to October	0.20	0.01	2.72	0.21
August to October	5.84	0.99	34.4	0.05
September to October	1.38	0.25	7.53	0.69
Temperature (°C)	1.44	0.94	2.21	0.09
Turbidity (NTU)	0.34	0.004	29.7	0.63
Total chlorine (0.1 mg/L)	1.77	1.26	2.50	0.001
Free chlorine (0.1 mg/L)	1.48	1.10	1.99	0.01
Location: house
cistern vs tap	1.42	1.02	1.97	0.04
Sample month:				< 0.001
July to October	1.07	0.54	2.14	0.85
August to October	4.42	2.24	8.70	< 0.001
September to October	0.63	0.32	1.21	0.16
Temperature (°C)	1.06	0.94	1.21	0.34
Turbidity (NTU)	1.00	0.93	1.08	0.95
Total chlorine (mg/L)	0.68	0.38	1.22	0.19
Free chlorine (mg/L)	0.86	0.45	1.63	0.64

CI: confidence interval.

Table 6. Final multivariable model for factors associated with total coliforms (count/mL).

Risk factor	Odds ratio	95% CI		p
		Lower	Upper
Location: truck
Sample month:				0.13
July to October	0.66	0.03	13.6	0.77
August to October	7.18	0.99	51.8	0.05
September to October	2.29	0.33	15.9	0.38
Total chlorine (0.1 mg/L)	1.68	1.14	2.46	0.01
Location: house
Cistern vs tap	1.51	1.07	2.13	0.02
Sample month:				< 0.001
July to October	1.17	0.58	2.38	0.66
August to October	4.69	2.35	9.38	< 0.001
September to October	0.63	0.33	1.23	0.18

CI: confidence interval.

The factors identified for consideration in building the final multivariable model for the presence of total coliforms in household water samples included sample type (cistern vs tap), sample month and total chlorine concentrations (Table 5). In the final multivariable model for total coliforms in household water samples, total coliforms were more likely to be identified in cisterns than in tap water samples (OR 1.51) and were more likely to be identified in August than in October (OR 4.69) (Table 6). Samples collected in August were also more likely to have positive cultures for total coliforms than samples collected in July (OR 4.00, 95% CI 2.24 to 7.14, p < 0.001) or September (OR 7.46, 95% CI 4.10 to 13.6, p < 0.001).

Factors associated with total chlorine concentrations

In a series of bivariate models (Table 7), total chlorine concentrations varied among sample months for both truck and household samples, but there was no significant association with turbidity or sample type (cistern vs tap). In the final multivariable model for the truck samples, total chlorine concentrations were lower in July than in October (Table 8). However, for the household samples, chlorine concentrations were higher in October than in August and September. Similarly, concentrations were higher in July than in August (p = 0.04) or September (p = 0.0004).

Table 7. Unadjusted associations between potential risk factors, chlorine concentrations and turbidity.

	Difference	95% CI		p
		Lower	Upper
Total Cl (mg/L)
Location: truck
Sample month:				< 0.001
July to October	−0.319	−0.475	−0.164	0.001
August to October	0.012	−0.150	0.175	0.88
September to October	−0.099	−0.243	0.044	0.16
Turbidity (NTU)	−0.120	−0.538	0.298	0.57
Temperature (°C)	0.030	−0.014	0.075	0.18
Location: house
Cistern vs tap	0.005	−0.083	0.092	0.92
Sample month:				< 0.001
July to October	−0.020	−0.168	0.129	0.80
August to October	−0.158	−0.308	−0.009	0.04
September to October	−0.252	−0.395	−0.109	0.001
Turbidity (NTU)	0.008	−0.014	0.029	0.48
Free Cl (mg/L)
Location: truck
Sample month:				< 0.001
July to October	−0.382	−0.531	−0.233	< 0.001
August to October	−0.062	−0.220	0.095	0.41
September to October	−0.124	−0.266	0.018	0.08
Turbidity (NTU)	−0.420	−0.857	0.017	0.06
Temperature (°C)	0.019	−0.028	0.067	0.42
Location: house
Cistern vs tap	0.052	0.025	0.079	< 0.001
Sample month:				< 0.001
July to October	−0.103	−0.182	−0.023	0.01
August to October	−0.246	−0.321	−0.171	< 0.001
September to October	−0.248	−0.311	−0.184	< 0.001
Turbidity (NTU)	0.011	0.001	0.020	0.03
Turbidity (NTU)
Location: truck
Sample month:				< 0.001
July to October	0.158	0.079	0.236	0.001
August to October	0.007	−0.076	0.089	0.87
September to October	−0.006	−0.080	0.069	0.87
Location: house
Cistern vs tap	−0.256	−0.559	0.047	0.10
Sample month:				0.27
July to October	0.115	−0.562	0.792	0.74
August to October	−0.332	−1.004	0.340	0.33
September to October	−0.365	−0.985	0.255	0.25

CI: confidence interval.x

Table 8. Final multivariable model for factors associated with chlorine concentrations and turbidity.

Risk factor	Difference	95% CI		p
		Lower	Upper
Total Cl (mg/L)
Location: trucks
Sample month:				< 0.001
July to October	−0.319	−0.475	−0.164	0.001
August to October	0.012	−0.150	0.175	0.88
September to October	−0.099	−0.243	0.044	0.16
Location: house
Cistern vs tap	0.008	−0.079	0.094	0.86
Sample month:				< 0.001
July to October	−0.019	−0.168	0.130	0.80
August to October	−0.158	−0.307	−0.008	0.04
September to October	−0.252	−0.395	−0.109	0.001
Free Cl (mg/L)
Location: trucks
Sample month:				< 0.001
July to October	−0.382	−0.531	−0.233	< 0.001
August to October	−0.062	−0.220	0.095	0.41
September to October	−0.124	−0.266	0.018	0.08
Location: house
Sample month:				< 0.001
Sample type:				< 0.001
Type*sample month:				0.03
Cistern vs tap: July	0.009	−0.048	0.066	0.75
Cistern vs tap: August	0.064	0.008	0.121	0.02
Cistern vs tap: September	0.111	0.063	0.160	< 0.001
Cistern vs tap: October	0.012	−0.054	0.077	0.73
Cisterns: July to October	−0.036	−0.128	0.056	0.44
Cisterns: August to October	−0.153	−0.241	−0.065	0.001
Cisterns: September to October	−0.173	−0.248	−0.099	< 0.001
Taps: July to October	−0.034	−0.129	0.062	0.49
Taps: August to October	−0.206	−0.297	−0.114	< 0.001
Taps: September to October	−0.273	−0.353	−0.193	< 0.001
Turbidity (NTU)	0.010	0.001	0.018	0.03
Turbidity (NTU)
Location: trucks
Sample month:				< 0.001
July to October	0.157	0.079	0.236	0.001
August to October	0.001	−0.076	0.089	0.87
September to October	−0.006	−0.080	0.069	0.87
Location: house
Cistern vs tap	−0.244	−0.55	0.062	0.12
Sample month:				0.31
July to October	0.052	−0.630	0.733	0.88
August to October	−0.370	−1.043	0.304	0.28
September to October	−0.383	−1.003	0.237	0.22

CI: confidence interval.

Factors associated with free chlorine concentrations

In the bivariate models, both month of sample collection and sample turbidity were associated (p < 0.20) with free chlorine concentrations (Table 7), in addition to type of household sample. In the final multivariable model for the truck samples, only sample month was associated with free chlorine concentrations (Table 8). Concentrations in July were significantly lower than in August (p = 0.0003), September (p = 0.0006), or October (p < 0.0001). In the final multivariable model for the household samples, free chlorine concentrations were higher in cistern than in tap water samples only for the months of August and September (Table 8). Free chlorine concentrations were significantly lower in cistern and tap water samples in both August and September when compared to October (Table 8).

Factors associated with turbidity measurements

Turbidity varied between months for the samples collected from the trucks (Tables 7 and 8). Turbidity was significantly higher in the truck samples in July than in August (p = 0.0007), September (p = 0.0001) and October (p = 0.0007). Neither sample type (cistern vs tap) nor sample month was significantly associated with turbidity in the household water samples (Tables 7 and 8).

Results: Qualitative interviewing

Truck-to-cistern supply and suppliers

When asked about the truck-to-cistern operations, participants indicated issues with the technicians and the cisterns:

I’m not a big fan of the water hauling situation on our reserve because for the last four years when I came back to my reserve, I keep saying ‘Who’s regulating those water haulers? How come their hoses aren’t covered? How come my mom and dad asked something to me this morning about the cistern cleaning?’ I need to check into what kind of dollars are available for cistern cleaning. How many times are those things supposed to be cleaned a year? Once? Twice? I say if somebody was to come and check out water haulers, we’d be shut down right now and told ‘Clean that up and do the proper way to haul water’. (BOFN-2)

A lot of them [cisterns] aren’t cracked all the way through but there’s cracks in them. At least if it’s all one piece it won’t let any liquid through, right? (BOFN-3)

I have three types of water. Rain, and the cistern, and the drinking water – bottled water. I don’t use the cistern. I don’t trust it. (BOFN-21)

Last year there was so much water and [name of community member]’s cistern was actually getting flow from underground. So we were digging it out and clearing it and pumping it out but there’s not really much you can do because it’s so low lying. They fixed everything, like the tanks and the seals, like the rings that up, the collar. (BOFN-4)

We do see that some cisterns have, they’re cracked. Some are not cleaned regularly, some have just kids depositing waste in there: some garbage, whatever. (BOFN-1)

Participants indicated that the main concern with cisterns was that cleanings were not performed regularly, annually or using certified cleaners (11/23 interviews).

Interviewer:

When were they cleaned?

Participant:

Summertime, but not anymore, I don’t think we do, but the students used to do that, you know, make them scrub it off, but I still wouldn’t drink it, I still would buy the drinking water. Yeah, for normal use at home it was used, but not for consuming, there was just something about it, I don’t know. Here I’m fine, cause I know it’s piped in, not coming from a cistern, I’m okay. (BOFN-13)

Lids were cracked or unlocked, and design and landscape conditions influenced the safety of cisterns (installed low-lying in flood-prone areas). Repairs and replacement parts were not easily obtained and/or installed (9/23 interviews). One participant indicated that in previous years, untrained summer students were hired to go around the reserve and clean cisterns. Two participants indicated that they cleaned their cistern themselves every year because service providers were not available or consistent with cleaning. They indicated that cistern lids were not secured in the community.

Yes, I drink my water from the cistern but I never drink it from the tap. I boil it, cause I am so used to buying bottled water, I buy bottled water. I buy my water and I put them in the fridge. It’s alright cause I had it tested. It’s drinkable but I find that having my cistern close to the road, it’s dusty. I had a cover made for me to be put over and then it walked away by itself. (BOFN-19)

Participants described cistern size as too small for the number of people drawing from them in each household, even with the cisterns being filled twice a week (10/23). Other practices such as lids being left ajar continuously for easy refilling were mentioned (6/23). One participant indicated that each time the cisterns are filled, more risk is introduced due to insufficient care being taken with lid removal, disinfection of hoses, and maintenance of the trucks and cisterns themselves.

Participants also indicated that operational problems are common. With two contracting companies providing truck hauling service, each with four or five different trucks and personnel, there were inconsistencies in protocols. For example, one participant indicated that ‘people were just standing around smoking cigarettes while they’re cleaning it! Yeah. And then the hose is an issue too. They’re just flapping around in the dirt’ (BOFN-2). Another participant explained that local wells were often being filled by contracted water haulers although they were not supposed to fill them. Residents drawing from those wells were on long-term drinking water advisories.

One participant indicated that random water quality tests of 8–10 cisterns a day were conducted but the tester had not received formal training to do so:

Yeah, [they] haven’t really had any, the only really training [they] had was when the Health Canada guys come out and just showed [them] how to do the sampling and stuff like that. There was supposed to be training but then I don’t know, it must have fell apart I guess. (BOFN-3)

The participant indicated there was testing for coliforms and E. coli on site, and collection of samples occasionally to send to laboratories for other parameters. When parameters were exceeded, the participant indicated that residents were informed and the tester took another sample right away for confirmation before implementing a boil water advisory. The participant said the tester attended conferences with the Water Treatment Plant Officer (WTPO) once or twice a year.

Also, participants valued the potential that a Source Water Protection Plan could offer their water supply but felt that there were barriers to putting this in place. Their perceptions of the barriers were explained as follows:

We need to help First Nations get the Source Water Protection Plans. Help them understand the importance of protecting their water. They [government agencies] spend so much money into putting people through, water treatment workers and that. Well, those two people can’t save the reserve. It’s all the people, different things that aren’t happening. You know we have Public Works, us – Lands, we have our farmers. What are they putting into the land? Those are the things that you have to look at first. You can educate all the water treatment plant workers you want in the world but it has to come from source water protection and that’s the key thing. (BOFN-2)

Some of the problems…um, encounters that basically, is our own community members and making them understand and educating them that this is an issue that we have to address, and also, one of the encounters that I’ve seen is the fact that we had new leadership every three years…. So every three years there’s a new chief and council that we have to educate and that’s frustrating, and safe water, sometimes, is not a primary issue compared to potash development or oil and gas development. (BOFN-1)

In sum, participants in BOFN felt that there were problems along the truck-to-cistern chain, beginning with protecting source water, and then with technicians, cistern infrastructure, local governance and training of personnel.

Discussion

The truck-to-cistern system examined in this study presented several sources of risk for drinking-water consumers. While the drinking water was measured to be within normal parameters as it left the water treatment plant to be loaded in trucks for transport, the water decreased in disinfecting agent availability, and increased in turbidity, temperature and presence of coliforms, during transport via trucks. Further, the probability of detecting total coliforms and E. coli increased as the water was stored in cisterns and travelled through household pipes. Re-contamination seemed to be a risk factor in this context. This study is the first of its kind to sample treated water along the truck-to-cistern supply chain in a Saskatchewan reserve community. The results support findings in other contexts indicating that cistern systems present health risks due to the high vulnerability, and subsequent frequency, of contamination (Levangie 2009; Lebel and Reed 2010; Neegan Burnside Ltd. 2011a; Coleman et al. 2013; Wright et al. 2017). The current study indicated the month of cistern filling was a risk factor for cistern systems, alongside previous studies demonstrating poor maintenance and upkeep of cistern systems, lack of training, inadequate infrastructure, and reduced technical capacity in First Nations communities (e.g. Maal-Bared et al. 2008; Lebel and Reed 2010; McCullough and Farahbakhsh 2012; Wright et al. 2017). It is hypothesized that over the summer months, the opportunism used to gain water hauling services is increasing risks to community members. A further contributor is the social dimension of changing leadership and changing priorities in communities, which means cistern maintenance is not performed regularly.

The water sampling results indicate that no detectable populations of coliforms originated within the plant, but detectable coliforms are appearing in the system during transport (coliform counts, and reduced total chlorine values), and during storage. Though this work did not seek to identify the exact origins of the coliforms, others have suggested potential vectors of contamination for truck-to-cistern systems: biofilms in containers and pipes, regrowth within cisterns, atmospheric dust and/or soil entering systems when refillable or when not properly covered and sealed, and contaminated groundwater flowing into cracked cisterns (Bowers et al. 2011; Farenhorst et al. 2017; Wright et al. 2017). In addition, human fecal contamination and animal waste (feces, saliva) present further threats when flooding or improper cistern filling occurs (Burket 2016; Farenhorst et al. 2017). It is interesting to note that E. coli was not found in any of the truck samples but was found in cisterns and tap water. Further studies on the maintenance regimes for technicians in cistern and piped systems are needed to assess contributing factors; however, our results agree with Daley et al. (2014) who stress the importance of learning contextual factors such as local social and cultural beliefs when investigating water supply and health threats.

Overall, coliforms were more likely to be reported in tap samples than in cisterns. This indicates that household cross-contamination during transfer from local storage tanks, and household plumbing, present risks for water users. Free chlorine was also significantly lower in tap water than cisterns for the majority of August and September. One potential way to reduce risks would be to increase chlorine concentrations from the treatment plant; however, experience denotes that on reserve, there is a strong preference against excessive chlorine use (Basdeo and Bharadwaj 2013; Dupont et al. 2014).

Conclusions

We posit that climate and human factors contributed to our findings. The late summer months are generally warmer and windier (McGinn 2010), contributing to increased risks of contamination of cisterns from dust and other particulates when filling, and of heat-enhanced bacterial regrowth in truck tanks, storage and pipes. Summers are also periods where there is increased water demand and usage, meaning cisterns are refilled more often. During summer months, children may be playing around cistern/well sites, domestic and wild animals may be more active near source water or around cistern/well sites, and untrained technicians may be used in lieu of others who may be vacationing or conducting deliveries elsewhere. Qualitative results support the multiple factors listed above and confirm that Health Canada’s (2012b, 2012c) non-binding guiding documents and INAC’s (2010) Protocol for cisterns are not being implemented on this reserve. As a result, multiple recommendations emerge from this study. Local practitioners need to:

1.	Increase local awareness (of water users, haulers, water treatment plant operators and reserve leadership) regarding potential contamination vectors for truck-to-cistern water supply systems, to reduce risks of contamination;
2.	Increase awareness about the Health Canada (2012b) guiding document for cisterns among water treatment plant operators, and other officials on this reserve, as well as seek ways to increase awareness among contracted water hauling companies serving other reserve populations;
3.	Contract a certified company to conduct water deliveries on a regular basis over the months of higher risk, and keep records of certifications of water haulers;
4.	Identify baseline bacterial compositions on the reserve, and install real-time monitoring technology to quickly identify and react when detectable coliform levels threaten public health.

Researchers (academic and from Indigenous representative groups such as the Federation of Sovereign Indigenous Nations [FSIN]) need to:

5.	Examine truck-to-cistern systems on other reserves in Saskatchewan and other provinces to identify patterns among contamination sources and infrastructural deficiencies;
6.	Evaluate cistern and well maintenance programs over a variety of contexts to determine better design, management practices, and policy gaps.

Government and chief and council agents need to:

7.	Seek prompt increases in federal funding for, and capacity building among, local people to maintain the cisterns sufficiently;
8.	Consider co-creation of locally relevant certification programs for water haulers and reserve personnel on truck-to-cistern systems so that contextual factors are included in training.

In the long term, all should work together to:

9.	Investigate the feasibility of using piped distribution systems on Saskatchewan reserves more widely where possible;
10.	Examine the cultural appropriateness, and potential expansion, of Health Canada’s (2012b, 2012c) guidance documents and INAC’s (2010) Protocol for reserves in Saskatchewan;
11.	Examine lessons learned from other contexts and countries using truck-to-cistern systems for policies, infrastructure and other innovations to implement where appropriate with Indigenous people in Canada.

The study herein found that coliform counts in the truck-to-cistern water system at one Saskatchewan First Nations reserve increased in storage and household taps, and were most likely to be detected in August. The results indicate that there is a public health risk to people reliant on this water system, and similar systems across Saskatchewan and Canada. The development of a better guideline and management strategy for truck-to-cistern systems with contextual factors included is relevant and important in this setting for two reasons: first, to contribute to Canada’s reconciliation with Indigenous people and understanding of how local human factors contribute to water stewardship and management; and, second, better guidelines would reduce risks and help develop rural resilience as reserve communities are faced with pressures from climate change, which could affect their drinking water supplies.

This work is not without limitations. The interviews were conducted over 2 years and focused on people’s perceptions of water quality and its provision on reserve, which could have changed since then; however, in recent communications with reserve members, there is evidence that perceptions have remained consistent given the lack of change in water provision. Water quality sampling was conducted by a graduate student and limited to 4 months in 1 year on one reserve. Trucks used for water hauling changed regularly and opportunistically. This finding points to additional problems in consistent water provision because of the lack of a fully certified fleet of haulers with whom a reserve could build capacity and quality control. Qualitative interviewing was also used, which presents a number of biases (observer, researcher and social desirability). Ongoing interviewing and monitoring would help to uncover further contributing factors. Other factors such as education and socio-demographics that may play a role in water quality monitoring and management were not addressed in this study.

An integral piece of this research was discovering through interviews that little information exchange and capacity building aimed at empowering communities to manage their water was occurring. This baseline data collected from Beardy’s and Okemasis First Nation will contribute to foundational knowledge and cataloguing of practices that contribute to risks for truck-to-cistern water. With ongoing work in Saskatchewan First Nations communities, a risk assessment framework and tools for practitioners could be developed and applied to other communities both regionally and nationally. This work presents the first comprehensive assessment conducted of risks from coliforms posed by cisterns in a First Nation reserve in Saskatchewan, and reiterates findings from other locales (Farenhorst et al. 2017).

Funding

This work was supported by the Canadian Institute of Health Research 2010-11-08 [grant number Operating Grant: Population Health Intervention Re] and Saskatchewan Health Research Foundation [grant number Research Connections Grant].

Acknowledgements

We would like to thank our community partners for their engagement, participation and assistance with the design of the study, sample collection, data management, and sharing their knowledge and experience with the university team and more broadly toward the goal of improved drinking water safety in First Nations.