Investigating diversity of pathogenic microbes in commercial bait trade water

Introduction

With over 30 million anglers in the United States and Canada, and with many of them using live bait in the form small fish (USDI, 2011; DFO , 2012), there is a significant risk of invasive species introduction and spread through the commercial bait trade vector (Drake & Mandrak, 2014). This is particularly alarming when commercial bait retailers are contaminated with invasive fish, such as Goldfish (Carassius auratus), Round Goby (Neogobius melanostomus), Eurasian Rudd (Scardinius erythrophthalmus), and Silver Carp (Hypophthalmichthys molitrix) (Nathan et al., 2015). Invasion risk significantly increases when anglers dump unused bait and water into the lakes and rivers at the end of a fishing day (Drake & Mandrak, 2014). However, this angling behavior has the potential to introduce more than just invasive fish species into new areas. Smith et al. (2012) revealed the water in which ornamental fish are transported contains a unique biota of fish and human pathogens beyond those found in the fish themselves. Additionally, the diversity of pathogens within wild and cultured baitfish is well studied (Goodwin et al., 2004; Lowry & Smith, 2007). As such, it is reasonable to suspect that water which contains bait fish could also serve as a reservoir for pathogenic bacterial species. This raises the question, what pathogens are found in the bait bucket water?

Pathogens have the potential to be very damaging to human and wildlife health (Daszak, Cunningham & Hyatt, 2000) and economically costly (Jenkins, 2012). With respect to fisheries, pathogens can be spread between commercial operations and wild populations, resulting in costly damages. Such was the case with amplified sea lice densities from farmed Atlantic Salmon leading to the decline of native Coho and Pink Salmon populations in British Columbia, Canada (Krkošek et al., 2007; Krkošek et al., 2009; Krkošek et al., 2011). The damages become more acute when the pathogens in question are generalists and spread throughout a valuable fishery, such as with viral hemorrhagic septicemia virus (VHSV) spread throughout the Laurentian Great Lakes (Rothlisberger et al., 2010; Escobar et al., 2017). While VHSV is a known pathogen already in the Great Lakes region, the identities and impact of other pathogens are largely unknown.

In the summers of 2012 and 2013, we visited over 500 bait shops across the U.S. Great Lakes states of Minnesota, Wisconsin, Illinois, Indiana, Michigan, Ohio, Pennsylvania and New York to collect water samples from commercial bait tanks for use in environmental DNA (eDNA) screening of invasive species (Mahon, Nathan & Jerde, 2014; Nathan et al., 2015). Our hypothesis was that if invasive species are in a bait tank, then the water would contain DNA from sloughed tissue, cells, and organelles in the water, which could be filtered, extracted, and screened using molecular tools to detect the invasive species (Ficetola et al., 2008; Jerde et al., 2011; Simmons et al., 2016). Additional to our initial hypotheses, these extracted eDNA samples also contain DNA from all organisms in the water, including potential pathogens, similar to those evaluated by Smith et al. (2012) who screened water samples for pathogens in the water of imported, ornamental fish.

Here, we repurpose the eDNA samples previously collected from commercial bait shops in the search for Great Lakes invasive fish species and analyze them using similar methods to those employed by Smith et al. (2012) to document putative pathogens. Our goal was to identify pathogenic species present in the samples, compare the diversity and abudance of bait shop sourced pathogens to Great Lake sourced pathogens, and evaluate the potential threat of unique, bait sourced pathogens being spread in the Laurentian Great Lakes in a manner similar to that documented for invasive species.

Methods

Sample collection and DNA extraction

Two-liter water samples were collected from the bait holding tanks in commercial bait shops from each of the states in the Laurentian Great Lakes basin (Table 1; Fig. 1; for additional collection details see Nathan et al. (2015)). Samples were filtered through ∼1.5 µm glass fiber filter paper within 24 h of their collection. DNA was extracted from filtered samples using a MOBio PowerWater DNA Isolation Kit (MoBio Laboratories, Carlsbad, CA, USA) following manufacturer recommendations. All samples for this study were collected and analyzed using previously described quality assurance protocols (Mahon et al., 2010; Jerde et al., 2011; Mahon et al., 2013; Jerde et al., 2013; Nathan et al., 2015). Samples were chosen for analyses based on two factors: (a) available DNA remaining from previous studies (Mahon, Nathan & Jerde, 2014; Nathan et al., 2015) and (b) proportional number of samples (out of 96 total) based on availability from each Great Lakes state where samples were collected. Upon consideration of those factors, samples were then randomly chosen for inclusion in this study. Total number of samples used in this study are listed in Table 1 by collection location. Because our previous collection site numbers (i.e., bait shops) were not equal from all Great Lakes states, proportionally, some states had more samples included in this investigation than others (Table 1). We chose to restrict our location information for individual bait shops (names, street addresses) where sample collections were taken to the US state to conceal the identity of individual vendors.

Table 1:

Sample collection locations.

Collection location (by US state) and number of bait shop DNA samples used in the study.

State	Number of samples
IL	6
IN	7
MI	53
MN	4
NY	9
OH	7
PA	2
WI	8

DOI: 10.7717/peerj.5468/table-1

Along with investigating water samples from commercial bait vendors, we included sequence data collected from a location in northern Lake Michigan to provide a point of comparison between water samples collected in commercial bait shops and in natural Great Lakes ecosystems (Hengy et al., 2017; Fig. 1). Five samples from one collection site were included in this study to serve as a comparison to our bait shop sequencing data. Although the Lake Michigan samples likely do not represent the true pathogenic diversity in the entire Great Lakes region, we included the wild samples to provide a comparison to the potential differences between wild and bait sourced pathogens.

Results

Sequence data associated with this study are available on the MG-RAST database (Meyer et al., 2008) under accession numbers mgm4791794.3 –mgm4791986.3 and as referenced in Hengy et al. (2017). From our resulting sequence data of the V4 hypervariable region of the 16S rRNA gene (515f/806r), a total of 1,594,572 sequence reads (of 8,082,960 total assembled sequences) matched the nine targeted groups listed in Table 2. Of these, Flavobacterium was the most diverse (400 OTUs) and most abundant (1,173,491 sequences) taxonomic group of putative pathogens (Table 3). Additionally, Flavobacterium was present in all of the eDNA samples processed and sequenced as a part of this effort. Least common in our resulting data was Plesiomonas, with a total of 17 reads found in the water sample from a single commercial bait shop in Michigan and the resulting BLAST search matches were to an unknown strain. Further, our results found that both potential human and fish pathogens were present in water samples collected in the commercial bait trade (Table 3). While we restricted our study to pathogens in trade (see Smith et al., 2012), our data from bait shop samples did show the presence of three putative fecal indicator bacteria: Bacteroides (13 OTUs with 9828 total reads), Staphylococcus (one OTU with 56356 total reads), and Enterococcus (one OTU with 3706 total reads). Future investigations on these and other comparable datasets should investigate these organisms further.

Compared to bait samples, the five water samples collected from one location in northern Lake Michigan had only five of nine of our chosen target groups present when sequenced and analyzed in same fashion (Table 4). Additionally, numbers of OTUs for Lake Michigan targeted groups were significantly lower (ranging from 1-39 total OTUs; Table 4). The Chi-square test indicated the distribution of pathogens was different between Lake Michigan and Bait shops (p-value <0.001, d.f. = 8).

Discussion

In this study, we documented the presence of human and fish pathogens in commercial bait retailers in the Great Lakes region using genomic surveillance. Compared to a sample sourced from Lake Michigan, bait samples had higher counts and higher diversity of multiple groups of pathogens. Bait samples collected for this study even found the presence of human fecal indicator bacteria (Bacteroides, Staphylococcus, and Enterococcus). Given the number of recreational anglers that use live bait and potentially dispose of bait bucket water into aqutic systems, the bait trade represents a potential vector for introducing and spreading pathogens throughout the Great Lakes. Along with this, angler’s use of bait presents the possibility of contact with the bait tank water increasing risk of exposure to these potential pathogens. While the virulence of these organisms remains unknown, this still represents a distinct possibility of transfer, spread, and/or infection when they are present in the system.

While most any water sample, be it from the holding water of an ornamental fish (Smith et al., 2012) or a commercial bait shop, local pond, drinking water source, or from a Great Lake, is expected to have some pathogenic biota, clearly there is a difference in the distribution and diversity of pathogens contained within samples. Our dataset found no evidence of Vibrio, Campylobater, Francisella, or Pleasiomonas bacterial strains in the limited number of samples from Lake Michigan water we sampled and sequenced, yet they comprised 1.5% of the sequenced pathogens in bait shop sourced water. While admittedly rare in our bait shop samples, these pathogens may be thriving or at some ecological advantage in bait shops or they may be at undetectable levels in the natural system. In contrast, OTUs similar to Aeromonas pathogens were nearly four times more abundant (as a percentage) in bait shops than in Lake Michigan sourced water. This again demonstrates an important point about these data, where a high percentage, i.e., dominant, microbes in bait water are potential pathogens, and in our Lake Michigan sampling, dominant microbes were likely non-pathogens. A study by Fujimoto et al. (2016) on lake water did not find OTUs classified as Vibrio, Mycobacterium, Aeromonas, or Pleasiomonas but did find one OTU that can be classeifed as Campylobater and Francisella. However, these were very rare as the study defined a total of 158,900 OTUs (Fujimoto et al., 2016). Although, in this study, the wild sourced Lake Michigan collections (from five samples sequenced and included here), came from a relatively limited spatial extent, which likely represent a fraction of the true pathogenic diversity in the Great Lakes, the contrast between the wild and bait shop samples suggests a substantial variation between the two sources. Future investigations should include a wider sampling effort from throughout the basin to make a more direct comparison between these two sources of water samples.

Conclusions

Detection and identification of pathogens is not new to science with over 19,000 articles published on pathogen detection in the last 10 years (from 2008-2018; IS Web of Science, query on 3/15/18). Researchers have previously reported on pathogen identification from areas outside of water and the environment, including the food industry, defense, and clinical applications (Lazcka, Campo & Muñoz, 2007). Analytical methods for these industries include traditional screening procedures (culturing and colony counts) through molecular methods and biosensors (Lazcka, Campo & Muñoz, 2007). However, the role bait water plays in pathogen transmission is unclear.

In the United States, there are approximately 33 million people that participate in recreational fishing (16 years of age and older) that account for an industry of over $48 billion annually and approximately 828,000 jobs (Southwick, 2012). Within this, the access to and the use of live bait from commercial shops presents a need to ensure the safety of participants. Spread of these pathogens is an ongoing important problem in the bait industry. Despite the concern over pathogens in the industry, actions towards prevention of pathogen spread by commercial vendors do not always address the issue (Connelly et al., 2018).

In this study, we note a number of potentially harmful pathogens in samples fully accessible to the public. However, there are a number of caveats to this. First, potential pathogens, albeit present in the samples as noted, could be at low levels and may never become virulent. Additionally, the pathogenicity of the documented OTUs was not specifically tested in this study. Second, there is no guarantee of spread from source (e.g., commercial shop) to additional sites, or that bait water bacteria could become harmful to humans as the pathogens may become damaged during transport or die when released into the environment, a similar observation to that of the importance of propagule pressure in the invasive species literature (Lockwood, Cassey & Blackburn, 2005). However, while there are no guarantees for spread and/or infection, the likely repeated introduction of these potentially dangerous strains of organisms to the environment and as documented in the invasive species literature, even low probability survival and arrival can ultimately lead to establishment and damages of invasive species (Jerde & Lewis, 2007; Jerde, Bampfylde & Lewis, 2009). Future screening and monitoring is needed to ensure safety for millions of participants in recreational fishing annually and also to the ecosystems where these harmful pathogens could be spread.