INTRODUCTION

Large-scale mortality events and declines in mussel populations have occurred with increasing frequency in recent decades (Strayer et al. 2004). Clear explanations, such as toxic spills, have been identified in some cases; in others, disease has been suspected, but not confirmed (Neves 1987; Haag 2019). Beginning in summer 2016, biologists observed a mass mortality event affecting numerous mussel species in the Virginia and Tennessee portions of the Clinch River. Subsequent investigations revealed that mortality recurred seasonally from late summer to early autumn. Although many species were observed dead or moribund, the Pheasantshell (Actinonaias pectorosa) mussel was affected disproportionately. Pheasantshell initially was among the most abundant species in the Clinch River, but population sizes declined 50–80% across multiple sites after mortality events (Richard 2018). In response to the mortality event in the Clinch River and a contemporaneous multispecies mortality event in Big Darby Creek, Ohio, USA, a research group was formed to study the events and gather baseline data to identify potential pathogens (Leis et al. 2018). The group reported a picorna-like virus from a Wabash Pigtoe (Fusconaia flava) in the upper Mississippi River (Goldberg et al. 2019); 17 novel viruses, including a densovirus associated with moribund mussels in the Clinch River (Richard et al. 2020); and a novel gonadotropic microsporidian (Knowles et al. 2022). They also conducted molecular and culture-based evaluations of the bacterial composition of mussel hemolymph from several river systems in the eastern United States (Leis et al. 2019; Richard et al. 2021).

Figure 1.

Locations where hemolymph samples were collected from mussels in the Clinch River, USA. Inset map shows location of the study area in southwestern Virginia and northeastern Tennessee.

In a previous study, we examined culturable bacteria associated with a 2017 mussel mortality event in the Clinch River (Leis et al. 2019). We identified many bacterial genera, but only Yokenella regensburgei was detected with high prevalence in Pheasantshells while mortalities were occurring, and it was not present a few months later after mortality subsided. This bacterium was previously identified from a mussel mortality event in the Tennessee River (Starliper et al. 2011), but whether it plays a direct role in such events remains unknown. Since 2017, episodic mortality of Pheasantshells has continued each autumn in the Clinch River. We investigated bacterial communities in the Clinch River during mussel mortality events in 2018, 2019, and 2020 and examined the spatial and temporal prevalence of bacterial genera among Pheasantshell and other unionid species.

METHODS

We collected samples from live and moribund mussels at seven sites in the Clinch River in 2018, 2019, and 2020 (Fig. 1 and Table 1). After observing mussel mortality in autumn 2016 and 2017, we established a series of sampling sites within and upstream of the zone of observed mortality and began sampling in summer 2018. We sampled six sites monthly from August to October 2018. High rainfall forced us to abandon planned sampling events in November and December 2018. In 2018, we sampled Pheasantshell and Mucket (Actinonaias ligamentina); the annual Pheasantshell mortality event began in September and no moribund Muckets were observed (Table 1). In 2019, we observed a mortality event that began in September and sampling occurred at Sycamore Island while the event was ongoing in October. We sampled moribund Pheasantshells and apparently healthy individuals of Mucket, Pocketbook (Lampsilis ovata), Three-ridge (Amblema plicata), Kidneyshell (Ptychobranchus fasciolaris), Wavyrayed Lampmussel (Lampsilis fasciola), and Purple Wartyback (Cyclonaias tuberculata). We observed mortality in October 2020 and collected targeted samples consisting of moribund Pheasantshells combined from three adjacent sites: Speers Ferry, Sycamore Island, and Clinchport. Later in the month, we also sampled moribund Pheasantshells, Muckets, and Cumberlandian Combshells (Epioblasma brevidens) from Sycamore Island.

Table 1.

Isolation and prevalence of Yokenella regensburgei in Clinch River, USA, mussels from 2018 to 2020. A. ligamentina = Actinonaias ligamentina; A. pectorosa = Actinonaias pectorosa; A. plicata = Amblema plicata; P. fasciolaris = Ptychobranchus fasciolaris; L. fasciola = Lampsilis fasciola; C. tuberculata = Cyclonaias tuberculata; E. brevidens = Epioblasma brevidens; L. ovata = Lampsilis ovata. N = number of individuals sampled.

In 2018 and 2019, we collected hemolymph from the anterior adductor muscle of each mussel by slightly opening the shell with a child nasal speculum, placing a stopper between the shells, and drawing out a hemolymph sample with a 1-mL syringe and 25-gauge needle. After collecting each sample, we immediately plated and streaked approximately 100 µL of hemolymph onto sterile tryptic soy agar culture plates (Becton Dickinson, Le Pont de Claix, France). Plates were shipped overnight to the U.S. Fish and Wildlife Service, La Crosse Fish Health Center, La Crosse, Wisconsin. We incubated the plates at 21°C for 7–14 d. After incubation, we used a sterile, disposable loop to remove morphologically unique colonies from each plate; placed them in a micro-centrifuge tube; and extracted DNA by using the PrepMan™ Ultra Sample Preparation Reagent (Thermo Fisher Scientific, Waltham, MA, USA). We subjected the extracted DNA to 16S rRNA gene PCR by using the same primers used by Leis et al. (2019), followed by Sanger sequencing (Eton Biosciences, Union, NJ, USA). We then edited and assembled the sequences de novo by using the default parameters in Geneious v11.1.5 ( https://www.geneious.com/download/previous-versions/#geneious-r11-dot-1 [accessed August 19, 2022]), and we identified resulting contig sequences through megaBLAST searches in the National Center for Biotechnology Information database ( https://blast.ncbi.nlm.nih.gov/Blast. cgi [accessed August 19, 2022]). In 2020, moribund mussels were wrapped in wet towels and sent on ice to the La Crosse Fish Health Center for processing as described above. Because Pheasantshell was the primary species observed in moribund condition, we used Fisher's exact tests to examine whether there were nonrandom associations between frequently observed bacterial genera and healthy or moribund Pheasantshell samples. For each bacterial genus present in six or more Pheasantshells, as well as for the condition of “no bacterial growth observed,” we set up a 2 × 2 contingency table with categories of bacteria presence/absence and healthy/moribund mussels. Pheasantshell samples within the moribund and healthy groups were pooled across all sites and dates from the study. The results of each Fisher's exact test indicate whether there was a statistically significant association between the presence of a particular bacterial genus and Pheasantshell health status.

Table 2.

Prevalence of the six most common bacterial genera and samples yielding no bacterial isolates in moribund and healthy Pheasantshell (Actinonaias pectorosa) mussels collected in Clinch River, USA, from 2018 to 2020. An asterisk (*) indicates statistically significant differences in prevalence between healthy and moribund mussels (Fisher's exact test: P ≤ 0.002). N = number of individuals sampled.

RESULTS

We examined a total of 91 mussels (67 Pheasantshells, 15 Muckets, 1 Cumberlandian Combshell, 1 Purple Wartyback, 2 Wavyrayed Lampmussels, 1 Kidneyshell, 2 Threeridges, and 2 Pocketbooks), including 49 healthy and 42 moribund individuals, from the Clinch River during 2018, 2019, and 2020. Bacteria were isolated from 80% (73 of 91) of the mussels sampled; 18 mussel samples yielded no bacterial isolates. All the cultured colonies were identifiable, except for two isolates from Muckets sampled on August 16, 2018, and October 25, 2018.

Across all sampling seasons, we identified 190 isolates belonging to 46 bacterial genera from 91 individual mussel hemolymph samples (49 apparently healthy, 42 moribund; Appendix A1). Most bacterial genera were observed only rarely, with 39 of the 46 genera present in three or fewer individual mussels and one present in four individuals (Appendix A1). The six most common genera identified were (in order of decreasing abundance) Yokenella, Bacillus, Microbacterium, Pseudomonas, Aeromonas, and Acinetobacter. The most common isolates for healthy mussels were Bacillus (27%; 13 of 49), Microbacterium (20%; 10 of 49), and Pseudomonas (16%; 8 of 49), with all other genera present in four or fewer individuals. The most common isolates for moribund mussels were Yokenella (57%; 24 of 42), Aeromonas (26%; 11 of 42), and Bacillus (14%; 6 of 42), with all other genera present in four or fewer individuals. Yokenella was observed in only three healthy individuals, whereas Aeromonas was never observed in healthy individuals. The prevalence of Yokenella and Aeromonas was significantly higher in moribund than healthy Pheasantshells (Fisher's exact test: P < 0.0001 and P = 0.0021, respectively; Table 2). The prevalence of the other four most common genera and the prevalence of samples yielding no bacterial isolates were not significantly different between moribund and healthy Pheasantshells (Table 2).

We observed Y. regensburgei each year during active mortality events in the Clinch River. Sequences identified as Y. regensburgei shared >99.3% similarity and were between 636 and 1,375 bp (Appendix A1). In 2018, Y. regensburgei was present in Pheasantshells at Speers Ferry, Sycamore Island, Wallen's Bend, and Kyle's Ford, all of which are sites where moribund mussels were observed (Table 1). The bacterium was not isolated from apparently healthy Muckets sampled at these sites or from any samples collected at Artrip, an upstream site where Pheasantshell mass morality has not been observed. All detections of Y. regensburgei in Pheasantshells occurred during periods of active mortality, except for one isolation from Wallen's Bend on August 16, 2018, which preceded our first observations of mortality by several weeks.

In 2019, Y. regensburgei was isolated from Pheasantshell, but not from six other mussel species; active mortality of Pheasantshells was also observed (Table 1). In 2020, during sampling that targeted moribund mussels, Y. regenburgei was isolated from 86% of Pheasantshells at three sites on October 7 and from 89% of Pheasantshells at Sycamore Island on October 20. Yokenella regensburgei also was isolated from moribund Muckets and Cumberlandian Combshell on October 20 (Table 1).

Aeromonas was detected only in 2020, when it was present in 11 of 18 moribund mussels collected. In one of these samples, two Aeromonas isolates were the only bacteria cultured, whereas in the remaining 10 samples containing Aeromonas, it co-occurred with Yokenella.

The prevalence of Bacillus spp. did not differ between apparently healthy mussels (44%; 16 of 36) and moribund mussels (17%; 6 of 36; Fisher's exact text: P = 0.1986).

DISCUSSION

The consistent association of Y. regensburgei with mussel mortality events and moribund mussels was one of the strongest and most conspicuous patterns of bacterial occurrence in our samples. We isolated Y. regensburgei, generally with high prevalence, during mortality events in every year of our study, and it was previously isolated during a mortality event in 2017 (Leis et al. 2019). Furthermore, it was rarely identified when mortality events were not occurring or at sites where mortality has not been observed (Artrip). The only occurrence of Y. regensburgei outside of a mortality event was its detection in an apparently healthy Pheasantshell on August 16, 2018, at Wallen's Bend; this may have represented an incipient occurrence at the onset of mussel mortality, which was observed a few weeks later at this site.

Yokenella regensburgei, along with predominantly Hafnia alvei, was identified from Ebonyshell (Reginaia ebenus) during mortality events in the Tennessee River, Alabama (2006 and 2008), and H. alvei was previously identified from the Clinch River (Starliper et al. 2008, 2011). Hafnia alvei and Y. regensburgei both are enteric bacteria that share similar biochemical characteristics, making separation of the two species uncertain by using traditional laboratory diagnostic methods (Lo et al. 2011). It is unclear whether Starliper et al. (2011) used molecular or biochemical techniques to identify Y. regensburgei and H. alvei. Furthermore, the Analytical Profile Index database (Biomérieux, Marcy-l'Étoile, France;  https://www.biomerieux-diagnostics.com/sites/clinic/files/9308960-002-gb-b-apiweb-booklet.pdf [accessed December 5, 2022]) used by Starliper et al. (2008) would have been unable to identify Y. regensburgei because that species is not included in the database, but H. alvei is included. Because of this limitation, it is possible that Y. regensburgei was present at higher prevalence during the Tennessee River mortality event. Our molecular methods should have allowed accurate separation of the two species, but neither we nor Richard et al. (2021) detected H. alvei in samples of mussel hemolymph from the Clinch River.

Despite the consistent association of Y. regensburgei with mussel mortality events, its role in these events is unclear. At least two scenarios could explain this association. The first scenario is that this bacterium is pathogenic. Preliminary histopathology work does not support pathogenicity (S. Knowles, unpublished data), but additional research is needed to confirm this result. The second scenario is that Y. regensburgei opportunistically colonizes mussels that are stressed and of compromised health due to a separate insult, such as exposure to environmental toxins or degraded water quality (see Leis et al. 2019). Richard et al. (2021) found a shift in bacterial communities of mussel hemolymph when mussels exhibit signs of apparent disease. An important question is whether there is a relationship between Y. regensburgei and Clinch densovirus 1 or other viruses identified from the Clinch River (Richard et al. 2020). For example, are these organisms pathogenic, or does a separate environmental factor (e.g., toxins, thermal stress, changes in water chemistry or algal communities) result in an immunocompromised state that allows unchecked bacterial growth and viral replication? Another important question is whether Y. regensburgei is consistently associated with mussel mortality events in other watersheds. Future work evaluating the importance of this bacterium would involve the development of a diagnostic assay to rapidly identify Y. regensburgei in mussels, which could also be used to search for potential environmental sources or reservoirs and to better understand the seasonality of its occurrence. In addition, in vivo infection trials are needed to evaluate the pathogenicity of Y. regensburgei to Pheasantshell and other mussel species, alone and in combination with other factors.

Although the prevalence of Bacillus spp. did not differ significantly between healthy and moribund mussels, there was a suggestive trend of higher prevalence in healthy mussels, a trend also noted by Leis et al. (2019). Members of Bacillus have several characteristics that, hypothetically, could be considered beneficial to freshwater mussels (see Leis et al. 2019). The lack of a significant difference in the prevalence of Bacillus between healthy and moribund mussels could be due to the persistence of these bacteria in moribund mussels after the onset of disease. Additional studies are needed to evaluate potential associations of Bacillus spp. with mussel health.

The strong pattern of co-occurrence between Aeromonas and Yokenella in 2020 is intriguing because it also was observed by Richard et al. (2021) (their study included samples from Clinch River mussels in 2017–2018) and Leis et al. (2019) (their study included samples from Clinch River mussels in 2017). Both studies found high Aeromonas spp. and Yokenella prevalence associated with moribund mussels from mortality sites, and the two genera often co-occurred in samples. However, Richard et al. (2021) found high Aeromonas spp. prevalence in 2018 samples from Clinch River mussels, whereas we observed Aeromonas spp. only in samples collected in 2020. It is possible these discrepancies are due to differences between metagenomics and culture-based techniques, differences in sampling strategy, or other factors. Gill et al. (2022) observed an increase in potentially pathogenic Aeromonas in gut samples from Plain Pocketbook (Lampsilis cardium) after experimental exposure to mixed agricultural contaminants. It is possible that the Aeromonas represents late-stage opportunistic infections of individuals previously stressed by pathogens, contaminants, or other stressors. Future field studies and experimental infection challenges would aid our understanding of the role of these bacteria in mussel mortality events.

DATA AVAILABILITY STATEMENT

Data for this study are available in Leis et al. (2022) ( https://doi.org/10.5066/P9SARYP3 [accessed December 5, 2022]).