Environmental Risk and Recovery

Publication Date: 2019

Abstract

Risks associated with flooding often persist after the water recedes. Although these risks are sometimes sensed by residents (seen, smelled, touched, or heard), they are rarely documented or measured. Over time, these risks may become “masked” by other factors. Furthermore, environmental testing, often used to assess post-disaster environmental impacts, tends to ignore ecological complexity and social history, perpetuating an incomplete understanding of one’s environment and risk. As such, the development of new knowledge systems is needed to unmask these risks so they can be incorporated into recovery efforts. Engaging disaster survivors in citizen science may serve to fulfill this need by increasing both social awareness and scientific knowledge of risk. We collected survey, soil, and interview data in Lumberton, North Carolina, following Hurricane Florence to examine the presence of risk due to fecal contamination of soil and the impact of citizen science engagement on recovery knowledge systems. Results indicate that citizen science served as an effective approach for facilitating discussions about environmental risk while contributing to a science-based dialogue about E. coli presence and persistence in residential soils post flooding. The data serves as an important proof of concept for future studies that aim to deploy citizen science after disaster.

Introduction

Natural hazards produce a variety of risks that interact in complex ways with social and environmental change during disaster response and recovery processes. Some of these risks are readily perceived by residents, but others may only be suspected, and still others are unknown entirely or may become masked over time. Recent disaster research draws attention to these hard-to-detect risks; yet many go undocumented, and currently little is known about the socioenvironmental processes through which risks may become masked or unmasked in the post-disaster context. In this pilot study, we use citizen science as an entry point to document the persistence of fecal contamination in soil—a potentially common but understudied environmental hazard associated with flooding events—and examine how environmental risks may be concealed or revealed following a disaster.

The extent to which citizen participation in the scientific process, also known as citizen science, can engage residents in disaster recovery, while increasing knowledge and awareness about local risks is currently unknown. We hypothesize that citizen science can serve as a valuable tool for building resilient communities, which are able to adapt and “bounce forward” following natural disruptions, by drawing attention to masked environmental and health risks that may be overlooked by traditional (expert) knowledge systems. This is especially crucial as communities throughout the world are grappling with increasingly frequent, severe, and unpredictable extreme weather events as a result of climate change (National Academy of Sciences, Engineering, and Medicine, 2016¹). Understanding how these hazard events affect local communities and environments requires knowledge production practices capable of addressing the inherent complexity and uncertainty of socioenvironmental systems. Citizen science that is sensitive to context and incorporates local knowledge with social and biophysical sciences offers a promising means of unmasking risks within post-disaster contexts.

In this study, we test the ability of citizen science to help build resilient and sustainable communities after disasters using pilot data collected in Lumberton, North Carolina, following Hurricane Florence in September 2018. We collect three primary sources of data:

• Residential soil samples to test for the presence of the fecal indicator bacteria E. coli in areas inundated by floodwaters after Hurricane Florence,
• Survey data measuring individuals’ ability to cope following the hurricane, and
• Semi-structured interviews focused on participants’ disaster and recovery experiences, including environmental and health concerns.

Results from our preliminary analysis offer important lessons for citizen science engagement in disaster recovery, as well as the processes through which risks may be revealed or concealed in the post-disaster context.

Post-Disaster Masked Environmental Risk

Natural disasters are associated with numerous health impacts, such as immediate post-disaster injuries and increased rates of infectious disease and corresponding health conditions. Concerns related to flooding specifically include polymicrobial wound infections from pathogenic contamination by seawater, freshwater, and soil (Cook et al., 2008²); fungal infections from mold and other sources (Benedict & Park, 2014³; Brandt et al., 2006 ⁴); diarrheal disease (Ligon, 2006⁵); and other bacterial diseases that make their way into human systems as a result of compromised water systems (Casteel et al., 2006⁶; Pardue et al., 2005⁷). Left untreated, many of these diseases carry the risk of a more protracted outbreak from direct person-to-person transmission. This can be especially dangerous in contexts of displacement and poverty.

Despite many known risks, the complexity of post-disaster environments creates a number of challenges for understanding how “riskscapes” (Müller-Mahn & Everts, 2013⁸) may be rearranged and reconstituted in the aftermath of extreme weather events. Disruptions within both social and ecological systems during and following disasters produce rapid change that has the potential to reveal or conceal existing environmental threats. Hurricane Floyd, for instance, helped expose the threat industrial agriculture posed to waterways in eastern North Carolina after floodwaters breached dozens of waste lagoons (Edwards & Driscoll, 2014⁹; Smith, 2014¹⁰). Conversely, residential mobility and land redevelopment following disasters may obscure previously known risks (see Elliott & Frickel, 2015¹¹). Disasters may also produce new risks through changes in the natural and built environment, social practices, or socioeconomic status. Natural disasters frequently lead to releases of hazardous material into the environment (Young et al., 2004¹²). For example, the 2011 Tōhuko earthquake off Japan’s northeastern coast triggered the meltdown of the Fukushima Daiichi nuclear plant, exposing over 200,000 residents to radiation (Gill & Ritchie, 2018¹³). Changes to the natural landscape can also produce new risks, such as when hurricanes destroy sand dunes, thereby increasing vulnerability to coastal water intrusion (Roman-Rivera & Ellis, 2018¹⁴).

Many post-disaster risks may be difficult to detect, overlooked, or otherwise concealed over time by behavioral, social, or environmental forces. The characteristic chaos and complexity associated with disasters create a context in which individuals may be less likely to acknowledge certain risks or seek treatment for health symptoms. Particularly during the immediate aftermath, affected residents may prioritize other recovery efforts over seeking medical attention for seemingly minor symptoms. Some health conditions may also be misdiagnosed by medical professionals. For example, certain serious and potentially fatal fungal infections that are common in post-disaster contexts often appear clinically similar to bacterial infections, creating the possibility for misdiagnosis (Benedict & Park, 2014). In other instances, residents may be unable to identify the substances to which they were exposed. In New Orleans after Hurricane Katrina, disaster survivors and first responders expressed general concerns about “toxic gumbo” (Frickel & Vincent, 2007) and polluted waters (Pardue et al., 2005), but exactly which contaminants or combination of contaminants were in the waters was unclear. Such situations can limit the ability for individuals to take informed precautionary measures and for healthcare professionals to acquire crucial information to effectively treat individuals.

The transition from short- to long-term recovery may also mask or unmask risks. Once immediate short-term needs have been addressed, chaos managed, and the visibility of physical impacts reduced, individuals may have time to better assess peripheral and less prominent disaster impacts. At times, disasters can also shed light on long-standing issues in a community (Kim & Olshanky, 2015¹⁵; Smith, 2012¹⁶), which may become increasingly visible as normalcy and functionality return. However, rebuilding and stabilization may further detract community members from critically assessing their post-disaster vulnerabilities and risks. For example, residents may feel relief from the regained sense of normalcy, without any concurrent reduction in disaster vulnerability. This is evidenced in long-term health impacts in communities post disaster, which are often indirect, poorly understood, and overlooked (Cook et al., 2008).

The most common way we seek to understand environmental risk is through environmental testing. This form of testing is used to describe and understand biological, microbiological, and chemical features of the environment, helping to inform risk reduction decisions. Although necessary for understanding emergent riskscapes post disaster, traditional environmental testing can provide only limited insight into environmental hazards. Specifically, environmental testing as practiced by regulatory agencies and other expert systems tends to ignore ecological complexity and social history in favor of conclusive (i.e., statistically significant) and universal knowledge. For example, following Hurricane Katrina, the U.S. Environmental Protection Agency determined flood sediments did not pose a public health risk in New Orleans. Yet, disciplinary standards and institutional mandates resulted in testing that missed former industrial sites with likely legacy contamination (Frickel, 2008¹⁷). Official testing also “produced little or no place-specific knowledge about contaminants in residents’ back yards or inside their homes” (Frickel & Vincent, 2011, p. 22¹⁸). This shows how formal science can generate knowledge of some hazards, while simultaneously masking others, and suggests the need for knowledge systems better suited for assessing the lived experiences of those impacted after a disaster.

Citizen Science as a Tool to Unmask Environmental Risk

The concept of resilience refers to the ability of communities to adapt and “bounce forward” following disasters or other disruptions (Adger et al. 2005¹⁹, ; Kendra et al., 2018²⁰). Understanding risk represents an important dimension of resilience (Weichselgartner & Pigeon, 2015²¹). Risk knowledge increases communities’ adaptive capacity by enabling them to reduce vulnerability to hazards and better prepare and respond to future hazard events (Cutter et al. 2008²²). However, the difficulty of detecting post-disaster hazards and complexity of socio-environmental interactions pose many challenges to local risk understanding. To address these challenges, scholars have called for scientific models that integrate place-based biophysical and social data (Cutter, 2003²³; Donovan & Oppenheimer, 2015²⁴), as well as participatory approaches able to incorporate local knowledge and engage the public (Mercer et al., 2008²⁵; Mercer et al. 2010²⁶). We hypothesize that citizen science is particularly well-suited for this task, and may contribute to community resilience by raising awareness of the complex nature of risk, thus enhancing both individual and community capacity to recover and prepare for future disasters.

Citizen science refers to the engagement of volunteer citizens in the scientific process (Silvertown, 2009²⁷). In more simple forms of engagement, citizens may only participate in data collection. At the other end of the spectrum, there is “extreme citizen science,” which involves citizens in all phases of the scientific process (i.e., from defining the problem to reporting the results) (Paul et al., 2018²⁸). The limited scholarship on the use of citizen science in disaster recovery efforts suggests a number of benefits for communities, including improved participation in disaster recovery efforts, increased community “buy-in” to disaster recovery efforts, detection of a broader range of impacts from disaster, offered legitimacy for activist claims in government policy, facilitation of local knowledge creation and ownership, and initiation and reinforcement of desirable feedbacks (McCormick, 2012²⁹; Tidball & Krasny, 2012³⁰; Paul et al., 2018). Citizen science may also help unmask post-disaster hazards by integrating social and biophysical data (Crain et al., 2014³¹), as well as promoting scientific literacy (Bonney et al., 2009³²).

In this pilot study, we focus on the persistence of fecal contamination in the soil as a result of flooding. In prior studies, researchers found “sporadic” levels of E. coli, an indicator of fecal contamination, at beaches in the Florida Keys after Hurricane Irma in 2016; however, they acknowledged a limited understanding of sources and pathways of fecal pollution post disaster (Roca et al., 2019³³). Separately, multiple indicators of fecal contamination were tested, with test results varying by method, revealing current knowledge gaps in testing procedures for fecal contamination (Casteel et al., 2007³⁴). Others have found evidence of contamination in areas inundated by floodwaters, although not in statistically significant concentrations (Pardue et al., 2005) or with statistically significant differences from non-flooded areas (Sinigalliano et al., 2007³⁵). This highlights the currently limited scientific understanding of how certain environmental risks spread and persist in a post-disaster environment, a knowledge gap that can have direct consequences on the ways individuals respond and recover from disasters.

Research Questions

In this study, we seek to understand how the hurricane recovery process may mask or unmask environmental risk, as well as how citizen science may contribute to a greater understanding of masked environmental risks post disaster. The following research questions guided this project:

1. What is the distribution of E. coli in residential soils four plus months after flooding from Hurricane Florence?
2. How does engaging survivors in a citizen science soil testing protocol facilitate discussion and deliberation about less visible environmental risks?
3. What are the impacts of project engagement on participants’ recovery and feelings of empowerment?

Disaster Context and Study Site Description

This study was conducted in Lumberton, a town in southeastern North Carolina, approximately 80 miles from the coast. Lumberton is located in Robeson County and has a population of approximately 21,542 residents who racially identify primarily as non-Hispanic White (39 percent), non-Hispanic Black (37 percent), and American Indian (13 percent) (U.S. Census Bureau, 2010³⁶). A substantial portion of the population lives at or below poverty levels (35 percent), with African Americans and American Indians disproportionately represented among poor and unemployed residents (Graettinger et al., 2018³⁷). Lumberton spans the Lumbee River (Lumber River), one of its primary recreational and cultural assets, but also a major source of flooding. A substantial portion of Lumberton lies within 100- and 500-year floodplains (Coastal Dynamics Lab, 2017³⁸), with high concentrations of Lumberton’s Black population and rental housing units in those areas (Graettinger et al., 2018).

Lumberton experienced serious impacts from both Hurricane Matthew in 2016 and Hurricane Florence in 2018. During Hurricane Matthew, the Lumbee River experienced historic flooding due to prolonged heavy rainfall, ultimately causing thousands of residents to evacuate and hundreds more to need rescue (Coastal Dynamics Lab, 2017; Graettinger et al., 2018). Hurricane Florence two years later likewise produced significant rainfall in Lumberton (over 21 inches). Florence made landfall near Wrightsville Beach, North Carolina, on September 14, 2018, as a slow-moving Category 1 storm that stalled over the Carolinas for three days, causing a record $22 billion in damage in North Carolina (National Hurricane Center, 2019³⁹).

Following both hurricanes, citizens and news outlets across North Carolina expressed concerns over potential fecal contamination as a result of flooded animal waste lagoons and other sources (e.g., Doran, 2018⁴⁰; Irfan, 2018⁴¹). Environmental concerns surrounding industrial agricultural, particularly hog production, are a long-standing and controversial issue in North Carolina, the second largest pork-producing state in the nation. Robeson County alone is home to approximately 127 animal feeding operations and 67 waste lagoons. At least 50 hog waste lagoons overflowed as a result of Hurricane Florence (Davis, 2018⁴²), an occurrence observed in preceding hurricanes as well (Casteel et al., 2006). During Florence, nearly two millions gallons of untreated sewage also spilled in Robeson County, adding to concerns of fecal contamination (National Hurricane Center, 2019).

Data and Methods

This study is informed by prior community-based participatory research through which residents defined the issues of concern and helped determine how the researchers could best aid in community-directed recovery. In an effort to encourage individuals to think about masked environmental risks, we engaged participants who experienced flooding from Hurricane Florence in a citizen science project centered on environmental testing for E. coli, an indicator of fecal contamination. The environmental testing was paired with semi-structured interviews and a pre-engagement hurricane coping self-efficacy survey to investigate the lived experiences of disaster and disaster recovery and to understand environmental risk perceptions and concern during long-term recovery from Hurricanes Matthew and Florence. Data resulting from the project includes E. coli test, survey, and interview data.

Following institutional review board approval of the study, participants were recruited primarily by local community specialists who advertised within the community and drew from their personal networks between January and March 2019. The study was promoted as a disaster recovery citizen science soil sampling project intended for individuals who experienced flooding from Hurricane Florence. Social media and in-person recruiting at a local city festival were also used to gather participants. Residents were eligible to participate if they experienced flooding from Hurricane Florence and lived in Lumberton. The resulting convenience sample consisted of individuals who expressed interest in participating. Residents had the option of participating only in the soil sampling or in the soil sampling, survey, and interview portions of the study.

A total of 20 participants completed the survey, soil sampling, and interview portions of the study. Of those who completed all three portions, 65 percent identified as White, 25 percent as Black or African American, and 10 percent as American Indian. The gender distribution was 55 percent female and 45 percent male. Most participants (90 percent) had some level of education beyond a high school degree (e.g., trade school, associate’s degree, bachelor’s degree, graduate degree) and were aged 60 years or older (70 percent). A total of 45 soil samples were collected from both public and residential areas around Lumberton for E. coli testing.

Hurricane Coping Self-Efficacy Survey

The hurricane coping self-efficacy survey is a seven-item measure designed to assess individual appraisals of coping following a hurricane (Benight, Ironson, and Durham, 1999⁴³). Demographic data (e.g., race, gender, age, and education) was also collected in the survey. The survey was given to participants before their citizen science training and engagement. The rationale behind the use of this measure was to gain a greater understanding of participants’ current coping self-efficacy in relation to hurricane recovery and to serve as baseline data from which to compare future engagement with participants (e.g., upon delivery of E. coli test results, future disaster events). Participants completed the hurricane coping self-efficacy survey by hand. Data from paper surveys were manually entered into an Excel spreadsheet. Mean and standard deviation values were calculated for each of the seven items.

E. coli Soil Testing

The soil sampling engaged participants to test their residential soil for E. coli. Participants were met at their homes and trained to collect their soil for E. coli testing. Before collecting samples, the researchers explained the project goals, sampling procedures, and purpose of the citizen science training. Participants then selected an area of their yard to sample (typically an area they associated with flooding) and sampled according to the training.

Soil samples were collected in sterile collection tubes, stored on ice, and transported to North Carolina State University to be processed within six hours of collection. An eluent was prepared from each soil sample using sterile water and processed using an IDEXX defined-substrate assay, a most probable number-based enumeration technique using Quanti-Trays 2000 and Colilert-18 media packs (Boehm et al., 2009⁴⁴; Byappanahalli et al., 2003⁴⁵; Pickering et al., 2012⁴⁶).

The following process was used to obtain estimates of E. coli concentration:

• To prepare the soil eluent, approximately 6 grams (+/- 0.25 grams) of soil were extracted from the collection tube and transferred to a 50-milliliter flat bottom conical tube using a sterile, disposable spatula.
• Forty milliliters of distilled water were then added to each conical tube, and the soil and water mixture was vortexed for 2 minutes.
• After settling for 2 minutes, 20 milliliters of the eluent were added to 80 milliliters of sterile water in a sterile dilution bottle, to acquire the 100-milliliter sample volume needed for the IDEXX assay.
• Following the addition of the Colilert-18 media, the trays were sealed and then incubated for 18 to 22 hours at 35 °C.
• Upon removal from the incubator, trays were analyzed using a black light chamber.
• Fluorescent wells, indicating the presence of E. coli in the well, were marked as positive.
• Based on the number of small and large wells fluorescing, the most probable number (MPN) of E. coli was determined (i.e., an estimate of the concentration of E. coli in the sample). The concentration of E. coli in the soil sample is reported based on the dry weight of soil. To determine the weight to dry-weight conversion, 10 grams of wet-weight soil were placed in a 105 °C oven for 24 hours.

To continue citizen engagement throughout the process, photos of the laboratory analysis along with brief descriptions of the procedures were mailed to participants (see Appendix A). This was intended as an educational intermediate product for participants while they waited for the soil results. Once all testing was complete, individual results were delivered to participants alongside an information pamphlet about E. coli (see Appendix B). A graph of community results was also provided alongside a comparison graph of E. coli presence on the North Carolina State University campus in an effort to avoid unnecessarily stigmatizing participants or their community. Results were explained to participants in person, followed by a short debriefing to gain insight into residents’ experience participating in the project. If participants could not be reached, the results were left at their residence, with researcher contact information.

Semi-Structured Interviews

The semi-structured interviews explored participants’ disaster, flooding, and recovery experiences from Hurricanes Matthew and Florence, as well as their current environmental and health concerns related to the disasters. The interviews started during the soil sampling portion of the citizen science project and concluded after the soil sampling procedure had been completed. Some participants chose to walk around their houses during the interview to better illustrate their points (e.g., flooding levels and locations). Most interviews were relatively short, averaging about 15 minutes, and ranging from 5 to 34 minutes. All interviews were audio recorded and transcribed verbatim. Interviews were open-coded by hand for emerging themes. Once the initial coding process was completed, interviewers convened to validate each other’s interpretations. Results provided are based on themes identified in this preliminary analysis. Next steps for interview analysis involve the development of a codebook based on these emergent themes and digital coding by two independent coders.

Results

Together, results from the survey, environmental testing, and interviews provide a glimpse into the complex nature of post-disaster environmental risk. Specifically, we show how risks may become masked or unmasked during the hurricane recovery process and suggest citizen science offers a promising means of both drawing attention to, and increasing our understanding of, hard-to-detect environmental risks.

Survey Results

The hurricane coping self-efficacy survey offers a snapshot into participants’ abilities to cope with stress and disruptions from the disasters and provides information to contextualize the interview results. Table 1 reports the mean scores for the 20 participants who completed all three portions of the study.

Table 1. Hurricane Coping Self-Efficacy Measures and Scores.

Survey results suggest that the participants were most confident in their abilities to nurture their relationships with family and loved ones in the context of disaster recovery. Participants also felt confident in their ability to protect themselves and their property post disaster. This finding may in part reflect the fact that most participants had stayed in their homes during Hurricane Florence, some with the primary intention to protect their homes. In contrast, participants’ scores on perceived ability to return to routine activities and deal with emotions experienced since the storm suggest more difficulties in these areas. Maintaining financial security and maintaining food and housing likewise exhibit lower averages among the indicators, suggesting that participants continued to struggle with financial difficulties and access to recovery resources four months post-hurricane.

Soil Results

A total of 45 soil tests were conducted for this project, including 20 from participants who completed all three possible portions of the study. Five of those participants had multiple soil samples from their home because of the size of their property. An additional 12 samples came from residents only interested in soil testing. Another eight public locations were selected based on participant feedback. E. coli was detected in 64 percent of total samples. For samples with E. coli present, concentrations ranged from 0.27 to 246.52 most probable number (MPN) per dry gram of soil and had a median value of 1.71 MPN per dry gram. While there are regulations for E. coli in drinking water and recreational waters, no standards currently exist for E. coli in soil from which to compare levels attained in this study.

Interview Results

Based on preliminary analysis of the interviews, we find that the disaster helped unmask some environmental risks while masking others; residents were most likely to discuss visible and easily perceptible risks; and participants suspected other, less perceptible risks, but discussed them in vague and general terms because of their lack of physical proof.

Disasters Mask Some Risks and Unmask Others

From the interviews, it emerged that disasters can unmask some risks while masking others. With regard to unmasking risks, physical destruction can serve to physically expose certain risks. For example, one elderly couple talked about how gutting their house during recovery revealed that their sewer lines were close to being compromised. As a result, they took the opportunity to address the risk before it caused any adverse health or environmental consequences.

The disaster itself also unmasked additional environmental risks, particularly related to geographic vulnerability to flooding. Most of the participants had very detailed and vivid descriptions of the flooding in their immediate vicinities during both Hurricane Mathew and Hurricane Florence—they recognized the source of the water, where the water flowed, and which homes were more or less vulnerable. In other words, experiencing flooding brought heightened awareness to localized flooding patterns. Some of the interviews revealed that this type of experiential knowledge, first gained during Hurricane Matthew, served as a resource to reduce participants’ vulnerability during Hurricane Florence. One participant, for example, explained how a neighbor was able to better prepare for the second hurricane:

One neighbor down the street, who we jointly did sandbags with, they did get flooding in Matthew. In Florence, he was able to build a little berm… and had a sump pump to keep him from flooding down there.

The ability to apply this type of risk knowledge, however, depended on financial and other resources, something that not all residents could easily access.

Other participants offered insight into how disasters may also conceal risks. For example, one resident emphasized how the mental and emotional toll of the disaster could detract from noticing health risks:

You’re already prone to a lot of stuff already. So you know, if you can get up and move and put your clothes on and eat a little bit, you’re not really paying attention because your mind is not focused on your health.

This quote illustrates the participant’s understanding about how risks can become masked, as well as how the process may compound existing vulnerabilities.

Prevalence of Visible Environmental Risks

Most risks that became unmasked, particularly during the initial phases of disaster, tended to be those that were easily perceptible and detectable to residents. During the interviews, most participants indicated that their lives had generally “returned to normal” in the 4-5 months since the hurricane and few expressed strong, ongoing concerns about environmental risks related to the storm. When asked about environmental or health concerns related to the hurricane, the most commonly and confidently discussed were physically visible risks. Mold, for example, was the most frequently cited concern, although none of the participants had ongoing mold issues in their homes at the time of the interview. In some instances, when asked about environmental concerns, participants thought that the interviewer was specifically referring to mold. Yet mold was also described as a normal problem in the area. As one former teacher who had mold allergies explained:

Everything in Lumberton has mold on it…I haven’t really done anything about [my allergies] but, you know, like [the] schools are full of mold to begin with around here. I used to teach and I sub now. So, I mean, mold is prevalent…it’s just, it’s warm and humid most of the time or just the climate is, so this is like a different world from Raleigh. Believe me.

Other environmental concerns cited by participants were the appearance of new wildlife, increased mosquito populations, and changes in the physical landscapes. For instances, several participants noted that the river and the water in swamp areas had yet to fully recede since the storm.

Although not a concern at the time of the interviews, some participants did discuss their concern about contaminated floodwaters. No specific source was referenced with certainty, but some individuals attributed the “smelly’ and “nasty” water to garbage, sewage, and oil. However, consistent with past research, as the floodwaters receded and life “returned to normal” concerns about contamination diminished (Frickel & Vincent, 2006). For example, a long-term resident and mother of two had the following exchange with an interviewer:

Participant: Everybody’s garbage just washed away everywhere. I wouldn’t let the kids go outside after Florence, even when the yard was dry.

Interviewer: And how long did that last?

Participant: I mean, until we went back to work and they went back to daycare.

In this particular circumstance, the participant remained suspicious of contamination even after the floodwaters had disappeared. However the quote suggests that once their lives began to return to normal, their concerns about contamination decreased without the physical or emotional reminders of disaster impacts.

Other participants suggested that “Mother Nature” or God would take responsibility for any environmental damage or hazards related to the storm. As explained by one participant, “And I figure the yard is taking care of itself… I’m sure Mother Nature gets rid of whatever it does.” This belief may help explain the reduced environmental concern as residents lives returned to normal. These findings suggest that collective knowledge and understanding of risk focuses primarily on the most perceptible—particularly visible—risks. Yet even concern over these prevalent risks began to wane as they became less visible and life returned to normal. That is, risks unmasked by the disaster were subsequently masked as the community transitioned from short- to long-term recovery.

Suspicions about Less Visible Environmental Risks

When participants did reveal concerns about less perceptible risks, they were often expressed as general worry without a specific or identifiable source. For example, individuals discussed concerns about water quality, damage to and contamination of personal goods, and impacts on wildlife and plants. As opposed to the physically visible concerns mentioned in the previous section, however, participants tended to speak more generally and with less confidence about these more concealed risks. For example, one participant raised concerns about water quality, saying, “My son’s a chemist. He said, ‘Don’t drink the water, mama.’” This participant did not precisely know why the water might not be safe but trusted her son (who she specifically identified as a scientist) and based her decision to buy and use bottled water on his advice. This quote also suggests a trust in science to reveal and understand environmental risks, particularly invisible risks. Another longtime resident with strong but vague health and environmental concerns echoed this point when he discussed the need for environmental testing:

It's been a great concern of mine because we have older folks that are suffering and sniffling, and they're trying to say whether it's the mold. But, no one has really did an air quality study or a soil study. This is the first time that someone has come to do those types of studies, which is vital to a community.

These examples suggest suspicions of environmental contamination or hazards that residents are unable to confirm without the aid of scientific expertise or testing. Another example of suspicion about less perceptible environmental risk highlights residents’ reliance on sensory cues and knowledge of their own personal level of vulnerability to inform risk decisions:

They were telling us, “You could save this.” But it had an odor. It had a smell, and had the stains on it, and it was wet. Instead of trying to keep that stuff, we just threw it away. Because I didn’t want any health issues because I was already battling, you know, I got breast cancer that same year. And then, I didn’t want any bacteria because I couldn’t take it.

Although the participant could not identify any specific contaminant, her suspicions that exposure could exacerbate existing health conditions prompted her to discard damaged personal items, despite others encouraging her to save them.

One final example illustrating suspicions about environment risk regards one participant who lost 400 to 500 pounds of her pond’s fish in the days following Hurricane Florence:

It was probably four days after the initial flood and a day after they sprayed for mosquitos. I came out in the morning… when I looked out, I noticed all this activity in the pond. And when I came out and walked out onto the deck, all the fish were just coming up and gasping for air. And then later on that evening and then the next morning, fish were floating all over the place, dead.

The participant suspected that the die-off was caused by chemicals sprayed for mosquitos or runoff from a nearby farm, but voiced uncertainty. Following the incident, the participant took proactive measures to understand the underlying cause by immediately taking a water sample to the local cooperative extension office. The office ended up misplacing the sample, leaving the participant in a state of “limbo,” uncertain how or when to move forward.

As illustrated in these three examples, suspicions about potential risks have real impacts on those who acknowledge their presence or potential presence. Unlike the visible environmental risks, individuals often do not have physical “proof” to attribute a cause. In some circumstances, individuals may still take precautionary measures to protect themselves, such as the participants who bought bottled water, threw out potentially contaminated items, and pursued environmental testing. In other instances however residents may simply disregard their suspicions, especially if they do not have the resources to address their suspicions. As one participant pointed out:

When you're ignorant of the fact, what precautions are you going to take? So, until someone soundly refutes and puts out the information, then there's only so much poor folks going to do anyhow.

Discussion

Knowledge of disaster-related risk may motivate individual and community action toward mitigating risk or appropriately responding to a realized risk condition. The environmental testing conducted as part of this citizen science project, and motivated by preexisting community concerns, shows evidence of being an effective entry point for facilitating discussions about disaster-related environmental risk, while also building rapport with residents and balancing insider-outsider power relations.

The results from the E. coli testing itself were important in contributing to a science-based dialogue about E. coli presence and persistence in residential soils post flooding, an area of research for which there is currently limited work. Because E. coli has a relatively short survival period in soil—generally only a few months in optimal circumstances (Maule, 2000⁴⁷) —and can come from multiple sources (e.g., animal fecal deposits, naturally occurring in the soil), the presence of E. coli in our study samples cannot be attributed to Florence flooding. However, the results provide baseline data for future testing in the event of subsequent flooding. Engaging disaster survivors in closing the existing knowledge gap can benefit not only scientific endeavors but may also promote a more holistic understanding of risk among participants while enhancing their capacity to mitigate, respond to, and recover from disaster disruptions related to environmental risk.

Next Steps and Future Work

The results presented in this report are preliminary, and further analysis is needed to better develop findings. Researchers are currently returning to households to deliver E. coli results to participants. Field notes based on responses and reactions to this engagement and Hurricane Coping Self-Efficacy surveys will be added to the data. This will allow us to begin to answer our third research question: What are the impacts of project engagement on participants’ recovery and feelings of empowerment? Insight from these initial inquiries may also lead to the development of a more standardized measure from which to evaluate the impact of citizen science projects on disaster recovery and adaptive capacity.

Furthermore, the baseline data of residential E. coli levels combined with a network of citizen scientists who are now trained to collect their soil for E. coli testing create an important infrastructure from which to base future quick-response work centered on unmasking post-disaster environmental risk. In the case of a future hurricane or severe flooding event in Lumberton, we would be able to contact those individuals who participated in this project to ask if they would be interested in re-testing their soil immediately post-flooding. As long as there was community interest, we would be able to efficiently deliver materials to residents who can collect their soil samples to be analyzed for E. coli. These data could then be compared to the baseline data gathered for the current project, providing a better opportunity to understand post-disaster changes to riskscapes.

Acknowledgements. This work was jointly supported by a grant from the University of Colorado Natural Hazards Center through their Quick Response Research program, which is funded by the National Science Foundation grant number CMM11635596, and a grant funded by North Carolina Sea Grant project R/18-RCE-3, as part of NOAA grant NA18OAR4170069.

We would like to thank our community specialists Angela Allen, Margaret Crites, Hannah Goins, Sallie McLean, and Nathan McMenamin, as well as laboratory assistants Sean Daly, Jason Frye, Jeremy Lowe, and Ozioma Nwachukwu for supporting this work. We would also like to thank the citizen scientists in Lumberton, North Carolina, who volunteered their time and contributed to the project.

References