Investigating the Ability of Edwardsiella ictaluri and Flavobacterium covae to Persist within Commercial Catfish Pond Sediments under Laboratory Conditions

Two prevalent bacterial diseases in catfish aquaculture are enteric septicemia of catfish and columnaris disease caused by Edwardsiella ictaluri and Flavobacterium covae, respectively. Chronic and recurring outbreaks of these bacterial pathogens result in significant economic losses for producers annually. Determining if these pathogens can persist within sediments of commercial ponds is paramount. Experimental persistence trials (PT) were conducted to evaluate the persistence of E. ictaluri and F. covae in pond sediments. Twelve test chambers containing 120 g of sterilized sediment from four commercial catfish ponds were inoculated with either E. ictaluri (S97-773) or F. covae (ALG-00-530) and filled with 8 L of disinfected water. At 1, 2, 4-, 6-, 8-, and 15-days post-inoculation, 1 g of sediment was removed, and colony-forming units (CFU) were enumerated on selective media using 6 × 6 drop plate methods. E. ictaluri population peaked on Day 3 at 6.4 ± 0.5 log10 CFU g−1. Correlation analysis revealed no correlation between the sediment physicochemical parameters and E. ictaluri log10 CFU g−1. However, no viable F. covae colonies were recovered after two PT attempts. Future studies to improve understanding of E. ictaluri pathogenesis and persistence, and potential F. covae persistence in pond bottom sediments are needed.


Introduction
The commercial catfish industry is one of the largest aquaculture industries in the United States and has consistently led all aquatic species in sales from 1988 to 2018 [1]. The professionals responsible for channel catfish (Ictalurus punctatus) and hybrid catfish [♀channel catfish (I. punctatus) × ♂blue catfish (I. furcatus)] production have experienced exceptional progress in the industry but have also had to deal with numerous and recurring challenges. One of the most costly and consistent issues that commercial catfish producers face are bacterial diseases [2], namely enteric septicemia of catfish (ESC) caused by Edwardsiella ictaluri [3] and columnaris disease (CD) caused by Flavobacterium covae [4].
Explanatorily, F. covae is a Gram-negative, aerobic, long filamentous rod-shaped, gliding, non-halophilic, yellow-pigmented, opportunistic pathogenic bacterium [4,[46][47][48][49]. Outbreaks of CD in commercial catfish species can occur via direct fish-to-fish transmission when a carrier sheds the bacterium or through the water column [50]. Pathogenesis of CD occurs during periods of high fish stress when temperatures and organic loads in ponds are high, fish are overstocked, and exposed to excessive handling [51]. While the specifics of pathogenesis are not fully understood, generally, the pathogenic bacterium first colonizes the host via attraction, adhesion, and aggregation mechanisms. This is followed by the production of endotoxins, exotoxins, and bacteriocins, which eventually lead to the pathogen overwhelming the host fish's immune system and causing the disease [52]. In addition, the mucus that naturally covers the gills and skin of catfish causes a more robust chemotactic response in F. covae (formerly F. columnare genomovar II) compared to F. columnare (formerly F. columnare genomovar I), indicating a potential relationship involving adhesion [53].
Economically, the CD has caused severe losses to the commercial catfish industry since 1922 [54]. It has been reported to be the second-highest disease observed on catfish farms in the United States [29]. Losses are estimated to be 30 million USD annually [55]. While average mortality due to CD is between 50-60%, ponds containing channel catfish fingerlings can experience up to 90% mortality [54]. In west Alabama, the highest number of fish losses were due to CD outbreaks, which equaled an estimated 14.6 million USD in foregone sales from 2015 to 2021 [2]. This discrepancy in economic loss and fish number loss is likely because fingerlings and young fish are the most susceptible to CD [56]. In similarity to E. ictaluri and virulent A. hydrophila (vAh) diseases, CD outbreaks and infections caused by F. covae can be chronic and recurring [48,[50][51][52]56,57]. Additionally, F. covae has several adaptations for the bacterium to survive and potentially persist in harsh environments.
Historically, F. covae growth has been most successful by using low-nutrient media [39,58]. Concerning growth and prevalence, CD can be influenced by increasing temperature, organic matter, and nitrite concentration in the water [52]. Like other aquatic pathogens, F. covae can form biofilms in aquaculture systems, with factors such as calcium concentration, temperature, hardness, salinity, and the presence of certain carbohydrates can impact the formation of biofilm and growth [4,31,46]. Cai et al. [46] reported that the optimal conditions for F. covae biofilm formation are at 28 • C, 360 ppm hardness, 5 ppt salinity, and when mannose is present. Shoemaker and LaFrentz [59] have reported the capability of F. covae to utilize fish mucus as a nutrient source, which may alter virulence and protein expression. Some Flavobacterium spp. can grow at temperatures as high as 45 • C, while most are considered psychrophilic or psychrotolerant [60]. Flavobacterium spp. have been found in numerous environments, including bodies of freshwater and seawater, sediments, soils, glaciers, ice, and freshwater shrimp and catfish ponds [60][61][62]. Adaptations for dealing with environmental stressors such as peroxide resistance, iron metabolism, heat shock proteins, and multiple stress response mechanisms have been found within F. columnare and F. covae genomes [44,63]. The bacterium can also cope with oxidative stress and prolonged starvation, and bacterial cells can be revived following starvation while expressing less virulence [64,65].
With the ability of the opportunistic pathogens E. ictaluri and F. covae to handle harsh environments, it is plausible that they may potentially be able to persist within commercial catfish ponds over extended periods. Sediments accumulate most rapidly in the first years of pond use and, on average, can accumulate as much as 40 cm of sediment over 15 years [66]. Sediments consist of inorganic and organic matter originating from biological sources, primarily phytoplankton, catfish wastes, and uneaten feed [66,67]. Because sediment and organic materials continue to accumulate on the pond bottoms, and the drastic changes within the pond during a production season allow many opportunities for E. ictaluri and F. covae to infect stressed fish [30] and, more importantly, provide a viable environment for the pathogens to persist. The primary focus of this study was to determine if E. ictaluri and F. covae can persist within submerged pond sediments while simultaneously observing how their populations change over time. Additionally, physiochemical components of the sediments were examined to determine if they correlated with observed population trends. We hypothesized that both E. ictaluri and F. covae would successfully propagate within this environment and that differences in population growth would occur between different sediment types.

Previous Study
Persistence trials (PTs) using isolates of E. ictaluri and F. covae were subjected to the same experimental conditions described [68]. The PTs' sediment samples, water, bacterial inoculum, and aquaria systems were prepared using the methods described below.

Experimental Design and System Preparation
All methods utilized for sediment sample collection, sediment, water disinfection techniques, and PT system preparation were the same as those described by Tuttle et al. [68]. In addition, the systems were in temperature-controlled lab spaces set to a targeted 27.5 and 27.0 • C for the E. ictaluri PT (EI PT ) and F. covae PT (FC PT ), respectively.

Bacterial Culture and Trial Preparation
The wild-type E. ictaluri isolate S97-773 (recovered from diseased channel catfish at the Thad Cochran National Warmwater Aquaculture Center in Stoneville, Mississippi; accession number: JX867005) was utilized for this study [14,69]. Cryopreserved S97-773 stocks were revived on brain-heart infusion (BHI) agar and incubated for 48 h at 28 • C. Next, a pure E. ictaluri colony was placed in 1 L of BHI broth and incubated at 28 • C and 115 revolutions per minute (RPM) for approximately 48 h. The broth culture was centrifuged at 4000× g for 10 min in a 5810 R benchtop centrifuge (Eppendorf North America Inc., Enfield, CT, USA), washed in cold 1X phosphate-buffered saline solution (PBS) with an adjusted pH of 7.4. Bacterial cells were resuspended and adjusted to an optical density of 0.200 ± 0.005 at 550 nm using an Eppendorf Biospectrometer ® Basic (Eppendorf North America Inc.), resulting in an average inoculum concentration of 8.33 × 10 7 colony forming units (CFU) mL −1 .
Preparation of the F. covae inoculum, using isolate ALG-00-530 (recovered from a diseased channel catfish at the Alabama Fish Farming Center in Greensboro, Alabama; accession number: MG516971), followed a similar procedure [70,71]. However, the culture media was modified by Shieh [58] containing the antibiotic tobramycin at a concentration of 1 mg L −1 of media (MST) resulting in a more selective media [72]. The F. covae was passed over the selective MST agar five times to ensure the bacterium had grown accustomed to the antibiotic. After the fifth pass, a pure colony of F. covae was placed into 1 L of Modified Shieh broth and incubated for 24 h at 28 • C and 115 RPM. Once the broth culture had grown, the bacterial cells were spun down, as mentioned previously, and instead washed with a 0.1X PBS solution with an adjusted pH of 7.0. Bacterial cells were resuspended and adjusted to an optical density of 0.200 ± 0.005 at 550 nm using a DR3900 visible spectrophotometer (Hach Company, Loveland, CO, USA) to determine bacterial concentration [73]. The final F. covae inoculum concentration in PBS was 1.78 × 10 7 CFU mL −1 .
A randomized block design was used for each PT to assign the four sediment types to the 12 total chambers. In each chamber, 20 mL of either E. ictaluri or F. covae optically adjusted bacterial inoculum was added to 200 g of sterilized sediment and 500 mL of disinfected dechlorinated city water. The sediment mixture was vigorously stirred with a Pathogens 2023, 12, 871 5 of 20 sterile stainless-steel spatula for 1 min durations every 5 min over 1 h. After the mixing period, water volume within each chamber was increased to a total of 8 L. To simulate production pond aeration, a 3.5 cm × 1 cm × 1 cm cuboid Pawfly air stone at a fixed location within each chamber would expel air supplied via a Whitewater Silent Air Pump™ v201 (Pentair Aquatic Eco-Systems™, Apopka, FL, USA) for 12 h beginning at 18:00 h and stopping at 06:00 h the following day.

Sampling and Bacterial Enumeration
Sediment in each chamber was collected and bacterial populations were evaluated, with sampling times as follows: 24 h post-inoculation (designated as Day 0), 48 h post (Day 1), four days post (Day 3), six days post (Day 5) and eight days post (Day 7), then every seven days following the fifth sampling. Cai et al. [32] described the methodology used to extract sediment and enumerate live colonies of S97-773 and ALG-00-530 for their respective trials. Approximately 1 g of sediment was collected from each chamber using a sterile 10-mL serological pipette, placed in a sterile 15-mL centrifuge tube, and centrifuged for 10 min at 667× g. Liquid supernatant was removed and the remaining sediment pellet (~1 g) was resuspended entirely in 0.1X PBS, creating a 1:10 mixture, and vortexed until the pellet was homogenized. Next, 250 µL of homogenized sediment solution was placed into six wells of the leftmost column of a 96-well plate and serially diluted (10-fold), as Chen et al. [74] described. Four serial dilutions of six 10 µL replicates were each plated onto E. ictaluri Medium (EIM) as the selective media [75].
The spread plate method [76] and MST media were utilized to enumerate ALG-00-530 colonies. Two technical replicate MST agar plates were used for the four targeted 10-fold serial dilutions [58]. The EIM and MST plates were incubated at 28 • C for 24 h. The plate counts of E. ictaluri and F. covae were recorded, and the final counts of CFU g −1 of sediment were determined using the appropriate correction factors. On each sampling day, viable E. ictaluri and F. covae colonies were picked, and both were cryopreserved in a 50% glycerol stock at −80 • C. Additionally, a representative colony underwent genomic DNA extraction for later polymerase chain reaction (PCR) confirmation. Any bacteria not confirmed to be the isolates of interest were designated as "unknown" and labeled as such, followed by their respective chamber name, sampling day, and PT.

DNA Extraction and PCR Confirmation
After colony enumeration, the colonies of the bacterial species of interest were picked and confirmed via polymerase chain reaction (PCR) protocols. Genomic DNA (gDNA) from all bacterial colonies was extracted using the EZNA â Bacterial DNA Kit (Omega Biotek Inc., Norcross, GA, USA). Finally, all concentrations and gDNA purity measurements were assessed using a NanoDrop™ One C spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA, USA).
For the E. ictaluri colonies, a 25 µL PCR reaction was constructed using 12.5 µL of Hot-Start Taq Master Mix 2X (Amresco LLC, Solon, OH, USA), 1 µL of ESCF and ESCR primers from an initial 10 µM stock solution [77], 75 ng of template gDNA, and nuclease-free (NF) H 2 O to volume. Thermal cycling was conducted using an Eppendorf Mastercycler ® X50s (Eppendorf North America Inc.). After optimization, thermal cycling parameters consisted of an initial denaturation at 94 • C for 3 min followed by 35 cycles of 94 • C for 30 s, 58 • C for 30 s, and 72 • C for 1 min, with a final extension at 72 • C for 5 min. Positive and negative controls were run in a thermal cycler with test isolates. Then, 5 µL of PCR product was separated on a 2.0% agarose gel, stained with SYBR Safe DNA Stain (Edvotek â , Washington, DC, USA), in a 1.0X Tris-acetate-EDTA running buffer using electrophoresis. All gels were run containing a positive control (S97-773), negative control (NF H 2 O), and a 50 bp DNA Step Ladder (Promega, Madison, WI, USA). PCR products were visualized using a VWR ® Real-Time Electrophoresis Systems LED transilluminator (VWR International, Radnor, PA, USA). To confirm F. covae colonies, 25 µL PCR reactions were constructed using 12.5 µL of Hot-Start Taq Master Mix 2X (Amresco LLC, Solon, OH, USA), 1.25 µL of FcFp and FcRp primers from an initial 10 µM stock solution [78], 75 ng of template gDNA, and NF H 2 O to volume. Optimized thermal cycling runs began with an initial denaturation of 95 • C for 5 min followed by 40 cycles of 94 • C for 30 s, 56 • C for 20 s, and 72 • C for 1 min, with a final extension at 72 • C for 10 min. Gel electrophoresis protocols were followed, as mentioned above.
Any colonies not PCR-confirmed as E. ictaluri or F. covae in their respective trials were designated as unknowns and labeled with the sampling day, chamber name, and PT. To accurately identify unknown bacterial colonies via the 16s rRNA gene, PCR products, and primers 63F and 1387R [79] were sent to Eurofins Genomics LLC (Louisville, KY, USA). After nucleotide base-pair results were trimmed and aligned in the Molecular Evolutionary Genetics Analysis (MEGA) software version 11 [80], base-pair sequences were inputted into the National Center for Biotechnology Information (NCBI) Basic Local Alignment Search Tool (BLAST) database [81].

Sediment, Water, and Statistical Analysis
All procedures for conducting water quality and sediment physiochemical analyses were the same as those described by Tuttle et al. [68].
A t-test was used to evaluate variances in sediment chemistry parameters. Changes in bacterial populations (log 10 CFU g −1 ) over time among four sediment types collected from two farms were determined using a two-way repeated measures analysis of variance test, with sediment type treated as a random blocking factor. Differences between overall populations within the two PTs were measured with a paired t-test. Significant differences were identified via a post hoc Tukey's Studentized Range-HSD. A correlation analysis was used to link soil chemistry parameters and bacterial populations (log 10 CFU g −1 ). Data from each physicochemical sediment variable were analyzed for normality using a Shapiro-Wilk test. If the results did not follow the normality assumption, Spearman's rank correlation was used. To control the false discovery rate, all multiple testing p-values for correlation analyses alone were adjusted using the Benjamini-Hochberg procedure [82]. A p < 0.05 was considered to be statistically significant.
Additionally, bacterial persistence curves (BPC) were created for each farm and each overall PT by fitting a smoothing spline (SS) model to the E. ictaluri and F. covae populations (log 10 CFU g −1 ; y-axis) at sampling days (x-axis) as previously described by Hussain et al. [83]. The smoothing parameter (λ) was selected based on the restricted maximum likelihood (REML) method [84] to balance both function smoothness and lack of fit. The fitted SS models were used to predict E. ictaluri and F. covae populations using an x-axis scale from 0-14 d with an interval of 0.001. For each BPC, 95% confidence intervals (95% CI) of predicted bacterial population curves were created via bootstrapping [85] using the boot package, version 1.3-28; [86]. Data were resampled with replacement 1000 times, with the SS model re-fitted to both the EI PT and FC PT population data. The 95% CIs were determined from the 2.5 and 97.5th percentiles. We considered the BPC descriptors to differ significantly between farms if their 95% CIs did not overlap. The G*Power 3.1.9.4 software was used for sample size calculations [87]. BPC analyses were conducted using R software, version 4.1.1 [88]. All other statistical analyses were performed with SAS ® version 9.4 [89]. All figures were plotted using SigmaPlot version 14.5 (Systat Software Inc., San Jose, CA, USA). All data were presented as the mean ± standard error of the mean (SE).

Edwardsiella ictaluri Persistence Trail
The temperature during this persistence trial remained at 27.5 ± 0.3 • C throughout the 14-day trial. Colonies of E. ictaluri began appearing on the selective EIM media on Day 0 (24 h post-inoculation). However, some E. ictaluri growth across different replicates experienced a lag period and did not begin appearing on the EIM agar until Day 1 (48 h post-inoculation). All test chambers were inoculated and mixed within the same period of 60 min, and there was congruent growth in all sediment types but not all replicates. Despite this, populations of E. ictaluri initially experienced a steady increase, followed by a moderate decline and plateauing pattern over 14 days (Figure 1).
All figures were plotted using SigmaPlot version 14.5 (Systat Software Inc., San Jose, CA, USA). All data were presented as the mean ± standard error of the mean (SE).

Edwardsiella ictaluri Persistence Trail
The temperature during this persistence trial remained at 27.5 ± 0.3 °C throughout the 14-day trial. Colonies of E. ictaluri began appearing on the selective EIM media on Day 0 (24 h post-inoculation). However, some E. ictaluri growth across different replicates experienced a lag period and did not begin appearing on the EIM agar until Day 1 (48 h post-inoculation). All test chambers were inoculated and mixed within the same period of 60 min, and there was congruent growth in all sediment types but not all replicates. Despite this, populations of E. ictaluri initially experienced a steady increase, followed by a moderate decline and plateauing pattern over 14 days (Figure 1).

Figure 1.
Persistence of Edwardsiella ictaluri S97-773 population (log10 CFU g −1 ) in sediment samples collected from 12 study chambers (2 farms × 2 ponds/farm × 3 replicate tanks per pond). Within each box plot, the horizontal line indicates the median, symbols indicate the mean and error bars around the symbol represent the standard error of the mean. Box plots with different lowercase letters are significantly different at p < 0.05.

Sampling and Bacterial Enumeration
Across all 12 test chambers, population numbers on Day 3 were significantly higher than those on all other days (t60 = 3.55, p = 0.0011), except for Day 5 (t60 = 2.93, p = 0.0516). On Days 5, 7, and 14, the E. ictaluri total population was not different according to the pairwise comparisons among those respective sampling days (p > 0.05; Figure 1). When comparing the sediment types from the two farms, the overall population of E. ictaluri in farm B sediment was similar to the total population of E. ictaluri in farm A sediments (Figure 2). On the smoothing splines encompassed by 95% CIs, containing the raw values of CFU g −1 and log10 transformed CFU g −1 values (Figure 3), there are no differences in any population peak, breadth, or range values between farm A and farm B sediments, or overall sediment counts (Table 1).

Figure 1.
Persistence of Edwardsiella ictaluri S97-773 population (log 10 CFU g −1 ) in sediment samples collected from 12 study chambers (2 farms × 2 ponds/farm × 3 replicate tanks per pond). Within each box plot, the horizontal line indicates the median, symbols indicate the mean and error bars around the symbol represent the standard error of the mean. Box plots with different lowercase letters are significantly different at p < 0.05.

Sampling and Bacterial Enumeration
Across all 12 test chambers, population numbers on Day 3 were significantly higher than those on all other days (t 60 = 3.55, p = 0.0011), except for Day 5 (t 60 = 2.93, p = 0.0516). On Days 5, 7, and 14, the E. ictaluri total population was not different according to the pairwise comparisons among those respective sampling days (p > 0.05; Figure 1). When comparing the sediment types from the two farms, the overall population of E. ictaluri in farm B sediment was similar to the total population of E. ictaluri in farm A sediments ( Figure 2). On the smoothing splines encompassed by 95% CIs, containing the raw values of CFU g −1 and log 10 transformed CFU g −1 values (Figure 3), there are no differences in any population peak, breadth, or range values between farm A and farm B sediments, or overall sediment counts (Table 1). Table 1. Mean, standard error (SE), and 95% confidence intervals (95% C.I.) for descriptors of smoothing spline models are presented in Figure 3.

Bacterial Isolate Genetic Confirmation
Unknown bacterial colonies first appeared in the sediment sourced from farm B on Day 1 (48 h post-inoculation) and were present in all sediment types by Day 3. The morphology and phenotypic expression of unknown bacterial colonies were more varied and diverse ( Figure 4). The first colony that appeared in the selective EIM (designated as colony type A) was confirmed to be S97-773 via PCR and 16s rRNA sequencing procedures. Colony type A was present on all sampling days throughout the trial (Figure 4). On sampling Day 3, other colonies appeared to have the same color and shape but were small punctiform and pulvinate (colony type C) or intermediate-sized (colony type E). Some colonies appeared to have nearly the same morphology as ones designated as colonies A and C but began to exhibit a translucent and erose margin at both large (colony type B) and smaller sizes (colony type D). On sampling Day 7, there were large colonies that expressed white/opaque (colony type G), dark green (colony type I), and yellow (colony type J) color morphologies. Finally, on sampling Day 14, colonies that exhibited a curled and seemingly dehydrated margin (colony type H) and a noticeably larger colony size with a lobate margin (colony type F) began appearing.   Table 1.  Table 1. Unknown bacterial colonies first appeared in the sediment sourced from farm B on Day 1 (48 h post-inoculation) and were present in all sediment types by Day 3. The morphology and phenotypic expression of unknown bacterial colonies were more varied and diverse (Figure 4). The first colony that appeared in the selective EIM (designated as colony type A) was confirmed to be S97-773 via PCR and 16s rRNA sequencing procedures. Colony type A was present on all sampling days throughout the trial (Figure 4). On sampling Day 3, other colonies appeared to have the same color and shape but were small punctiform and pulvinate (colony type C) or intermediate-sized (colony type E). Some colonies appeared to have nearly the same morphology as ones designated as colonies A and C but began to exhibit a translucent and erose margin at both large (colony type B) and smaller sizes (colony type D). On sampling Day 7, there were large colonies that expressed white/opaque (colony type G), dark green (colony type I), and yellow (colony type J) color morphologies. Finally, on sampling Day 14, colonies that exhibited a curled and seemingly dehydrated margin (colony type H) and a noticeably larger colony size with a lobate margin (colony type F) began appearing.  Table 2.   Table 2.
Although colonies more phenotypically varied in this PT, PCR product bands using the ESCF and ESCR primers resulted in all isolated colonies producing the same amplicon region (177 bp) consistent with the positive control, indicating no apparent differentiation between the isolated bacterial colonies ( Figure 5). A more robust confirmation was conducted, and the 16s rRNA sequencing revealed six different species not identified as E. ictaluri (Table 2). Additionally, it would appear that three distinct bacterial colonies that initially appeared to be different from the species of interest were identified as E. ictaluri.
Although colonies more phenotypically varied in this PT, PCR product bands using the ESCF and ESCR primers resulted in all isolated colonies producing the same amplicon region (177 bp) consistent with the positive control, indicating no apparent differentiation between the isolated bacterial colonies ( Figure 5). A more robust confirmation was conducted, and the 16s rRNA sequencing revealed six different species not identified as E. ictaluri (Table 2). Additionally, it would appear that three distinct bacterial colonies that initially appeared to be different from the species of interest were identified as E. ictaluri.

Sediment and Water Analysis
The water quality parameters did not noticeably fluctuate throughout the PT ( Table  3). The sediments used in this PT are the same as the four sediment types used in a previous study, and all physical and chemical parameters between the two farms were not different [68]. Due to the small sample size of sediment physiochemical properties, the correlation analysis indicated no correlation between E. ictaluri populations over time and the sediment parameters (Table 4). Power analysis revealed the sample size required to determine the statistically significant correlations between E. ictaluri populations and each

Sediment and Water Analysis
The water quality parameters did not noticeably fluctuate throughout the PT ( Table 3). The sediments used in this PT are the same as the four sediment types used in a previous study, and all physical and chemical parameters between the two farms were not different [68]. Due to the small sample size of sediment physiochemical properties, the correlation analysis indicated no correlation between E. ictaluri populations over time and the sediment parameters (Table 4). Power analysis revealed the sample size required to determine the statistically significant correlations between E. ictaluri populations and each sediment parameter (Table 4). After the bacterial enumeration procedure was complete, all sediment samples were frozen.   Table 4. Results from correlation analysis tests between log 10 CFU g −1 of Edwardsiella ictaluri and sediment physicochemical variables. Based on bivariate normality testing, Spearman's rank correlation (coefficient = ρ) was used. All raw p-values were adjusted using the Benjamini-Hochberg procedure to control the false discovery rate (FDR). Significant results at p < 0.05. The sample size (n) is required to reveal statistically significant correlations.

Flavobacterium covae Persistence Trial
This PT was conducted on two separate occasions following all procedures described above. The water temperatures for the first and second attempts were approximately 27.2 ± 1.2 and 27.0 ± 0.4 • C, respectively. In both instances, no colonies of F. covae were recovered from the sediment over seven days. Incubation times were increased to 72 h to ensure that F. covae colonies were given ample media contact and propagation time, however, no colonies of F. covae propagated. Due to the lack of F. covae colonies being recovered in either PT, none of the statistical analyses described previously were conducted. In both FC PT attempts, viable colonies of unknown bacteria appeared on sampling Day 1 (48 h post-inoculation), displaying various unique colony morphologies. However, 16s rRNA sequencing outputs revealed that none of the colonies were F. covae or any Flavobacterium spp., revealing 12 distinct species (Table 5). Similar to the EI PT , all sediment samples were frozen after sampling. Table 5. Results from the NCBI Blast database for nucleotide 16s rRNA sequences from isolates collected during FC PT and the sampling day durations the unique colonies were present. Bacterial species were determined to have the highest probability under percent max identity, highest total score, and highest max query cover to show the percentage of query DNA covered.

Discussion
The results of the EI PT indicate that E. ictaluri can persist within the submerged sediments of commercial catfish ponds in a controlled laboratory setting. The bacterial growth curve illustrated by the data indicated that E. ictaluri populations began to plateau by Day 5 and did not change throughout the remainder of the trial. The highest average population across all sediments were log 10 6.4 CFU g −1 . Due to no overall difference between sediments sourced from the two different farms, this suggests that sediment has no apparent influence on the growth and maintenance of pathogen populations. These findings are consistent with those Plumb and Quinlan [37] reported and displayed similarities to how vAh behaves under similar experimental conditions [68]. Like vAh, E. ictaluri populations experienced a growth period, followed by reaching the stationary phase, and then plateaued to remain at a consistent population. However, unlike vAh, the bacterial populations in the EI PT began plateauing by the fifth sampling day, compared to a vAh persistence trial when bacterial populations began to plateau by the fourteenth sampling day [68]. The implications of this study provide valuable insight into the ability of these pathogenic bacterial species to survive in atypical environments. These findings, as well as Tuttle et al. [68], indicate the plausibility of E. ictaluri and virulent A. hydrophila to survive within the sediments of operating commercial catfish ponds as well as potentially other aquatic environments. This study adds another layer of understanding and prompts future research to better understand bacterial pathogenesis within catfish species.
Aside from this study, there are very few publications with direct evidence regarding the ability of Edwardsiella spp., let alone E. ictaluri, to survive or persist within sediments or soils of aquatic environments. E. ictaluri-specific phages found in water and sediments in a river in Hiroshima Prefecture, Japan [90], have been linked to an individual forktail bullhead (Pelteobagrus nudiceps). Viable E. tarda colonies have been found in the sediments and water of aquaculture ponds, and genetic differentiation exists between isolates found exclusively in sediments versus isolates collected from other sources [91,92]. In addition, E. tarda found in the soils of Owerri, Nigeria, displayed potential as a species for bioremediation of crude oil [93]. Regarding sediment, the correlation analysis could not distinguish significant physical and chemical factors of the sediment that influenced the population of the bacterial pathogen. The power analysis revealed a larger sample size is necessary to determine statistical significance with high power. These sample size numbers, which were in the thousands, would be unrealistic and cost prohibitive within the scope of this study but would be worth future investigation. It is also necessary to determine which cellular mechanisms and virulence factors allow for the persistence of E. ictaluri in the sediments of catfish ponds.
It has been established that species of Edwardsiella are naturally resistant to colistin [75]; however, it is notable that the bacteria identified in this study exhibited colistin resistance consistent with previous research findings. Genera from the family Enterobacteriaceae, such as Salmonella spp., Klebsiella spp., Aeromonas spp., Citrobacter spp., Shigella spp., Enterobacter spp., and Escherichia spp. contain mobilized colistin resistance genes [94]. Clostridium perfringens, Bacillus subtilus, Neisseria meningitides, Burkholderia spp., Proteus mirabilis, Kluyvera spp., Cronobacter sakazakii, Raoultella ornithinolytica, and Pseudomonas aeruginosa have all shown some level of resistance to colistin [95][96][97][98]. The most interesting finding from this study comes from verifying the identities of diverse colonies that grew on the selective EIM.
Based on the results of the 16s sequencing, it is apparent that four bacterial colonies were expressing different pigmentation and sizes that were all determined to be E. ictaluri. One reason E. ictaluri was chosen for this study, as opposed to E. tarda or E. piscicida, is due to the reported high phenotypic and biophysical homogeneity among isolates [99,100]. During the beginning of the persistence trial, E. ictaluri colonies produced a green pigment on the EIM; however, some apparent E. ictaluri colonies began expressing a yellow pigment by the fifth sampling day. One study reported that E. tarda colony pigmentation was black when grown on Salmonella-Shigella agar [101]. Bacterial pigmentation is quite diverse, and all unique pigments have a specific function essential for bacterial survival and ecological success [102]. Different bacterial genera, including Pseudomonas spp., Janthinobacterium spp., Streptomyces spp., Nocardia spp., Thermomonospora spp., Microbispora spp., Streptosporangium spp., Rhodococcus spp., and Kitasatospora spp., have diverse pigmentation [103]; however, there have been no studies reporting one species or genus of bacteria being capable of expressing two different pigments. Johansen et al. [104] demonstrated through genetic modification of the motility, cell shape, stringent response, and tRNA modification genes of a Flavobacterium spp. strain Iridescent 1, which could alter the nanostructure, which resulted in multiple colors observed among the same bacterial colonies. They also suggested that the structural color of bacterial colonies may be linked to cellular functions and gene activity, which may have significant implications for natural populations of pathogenic bacteria. To date, no studies have determined the natural pigments produced by E. ictaluri or if there is a linkage between cellular functions and pigmentation. Whole genome sequencing of the isolates collected during the EI PT would reveal what pigments these bacteria can produce and add another level of confirmation to the identity of these bacteria.
Even though both attempts to propagate F. covae colonies within this experimental design were unsuccessful, these findings are intriguing. Multiple factors may have influenced the lack of F. covae in these laboratory persistence trials. Environmental conditions such as water hardness, high temperature, organic matter, and nitrite concentration can increase the adhesion and virulence of the bacterial pathogen [105][106][107]. The biofilm formation is most effective between 25-28 • C and can be inhibited when salinity is as low as 3 ppt and significantly reduced at salinities over 7.5 ppt [46,107]. Another factor could be due to ecological interactions and interspecific competition. Bacterial species including Bacillus subtilis, Luteimonas aestuarii, Rhodococcus qingshengii, Leucobacter luti, and Dietzia maris were antagonistic towards F. covae and F. psychrophilum [30,108,109]. Additionally, tannic acid can act as an effective bactericide for F. columnare and E. ictaluri [110]. It could be possible that some of the other bacteria that appeared in the FC PT and some natural compounds or ions in the sediment prevented the establishment of F. covae.
The culture conditions necessary for successful F. covae growth can be sensitive. Although previous studies have reported that the best growth of F. covae is on low nutrient media [58,73,111], and the bacteria are slow growing [112,113]. While Shieh media has typically allowed for fast and effective growth of Flavobacertium spp. [114], a recent study indicated that G media provides effective and uniform distribution of F. covae colonies within 24 h [113]. Other media, such as tryptone yeast extract salt media [112] and antibiotics such as polymyxin-neomycin have been utilized to create selective media [73] for successful F. covae growth. Media type and culture considerations for future persistence studies may yield more favorable results.
Since we know the pathogen F. covae and other members of the Flavobacterium genus have been found in aquatic environments outside of a host [50,[115][116][117][118], it is plausible that F. covae may have the ability to persist within the environment. In addition to biofilm formation, a recent study by Abdelhamed et al. [119] revealed that F. covae could grow under anaerobic conditions via denitrification genes and nitrite reduction. However, at this time, we were unable to verify the ability of the pathogen F. covae to persist within the sediments of commercial catfish ponds under the conditions outlined in this study.

Conclusions
Understanding the mechanisms that allow these pathogenic bacteria to persist within sediments is vital for effective disease management strategies for commercial catfish producers. E. ictaluri has been confirmed to be able to persist within aquatic sediments based on the results of this study, however, this may have profound implications. Further gene expression analyses such as those conducted on vAh [120] may reveal that cell origins of E. ictaluri may result in different virulence factor expressions. Additionally, E. ictaluri persisting within sediments may be more susceptible to developing anti-microbial resistance [121], as has been reported in previous studies [36,[122][123][124]. Conversely, since F. covae propagation was unsuccessful, modifications to this experimental design will be necessary for future studies. While F. covae can form biofilms, other environmental and experimental factors within the aquatic environment may contribute to them not being recovered in this study.  Institutional Review Board Statement: No animals were used in this study and IACUC approval was not required. All microbial work was carried out under a Biological Use Authorization (BUA #926) approved by the Auburn University Institutional Biosafety Committee (IBC).

Informed Consent Statement: Not applicable.
Data Availability Statement: All data from this study are available from the corresponding author upon reasonable request.