N M D S 1

Figure 5. The gut microbiomes of traditional hatchery-reared steelhead differ compositionally compared to their wild broodstock counterparts. NMDS plot showcasing the differences between hatchery broodstock and wild broodstock gut microbial samples (PERMANOVA R<sup>2</sup> = 0.29, p < 0.01) as well as differences between Cedar Creek Hatchery and North Fork Alsea Hatchery locations (PERMANOVA R<sup>2</sup> = 0.07, p < 0.001). Stress = 0.13. Visually, there is separation between wild steelhead gut microbial composition and hatchery steelhead gut microbial composition within their respective river basins. **Figure 5.** The gut microbiomes of traditional hatchery-reared steelhead differ compositionally compared to their wild broodstock counterparts. NMDS plot showcasing the differences between hatchery broodstock and wild broodstock gut microbial samples (PERMANOVA R<sup>2</sup> = 0.29, *p* < 0.01) as well as differences between Cedar Creek Hatchery and North Fork Alsea Hatchery locations (PERMANOVA R<sup>2</sup> = 0.07, *p* < 0.001). Stress = 0.13. Visually, there is separation between wild steelhead gut microbial composition and hatchery steelhead gut microbial composition within their respective river basins.

#### *3.3. Combination of Both Studies*

After combining all our available wild-born, wild broodstock, and hatchery broodstock gut microbiome samples, we found that the river basin, broodstock history, and weight remained associated with the beta diversity of the steelhead gut microbiome (PER-MANOVAbasin, Bray–Curtis, R<sup>2</sup> = 0.1223, *p* = 0.001; PERMANOVAbroodstock, Bray– Curtis, R<sup>2</sup> = 0.0479, *p* = 0.0001; PERMANOVAweight, Bray–Curtis, R<sup>2</sup> = 0.1170, *p* = 0.0001). Additionally, the dimensions displayed in the NMDS show two all-encompassing potential clusters that we confirmed with a partitioning around medoids (PAM) cluster-based analysis (Figure 8). Despite their separation, these clusters are not explained by any of our

measured variables, suggesting some other variables underlie this observed structure in the diversity of the steelhead gut microbiome. Microorganisms 2022, 10, x FOR PEER REVIEW 10 of 16 Microorganisms 2022, 10, x FOR PEER REVIEW 10 of 16

> Figure 6. Examples of gut bacterial clades that are more abundant in either traditional hatchery or wild broodstock fish. Asterisk (\*) indicates a statistically significant result using Kruskal-Wallis tests and false discovery rate multiple test correction. (A) This Peptonophilus clade example is more abundant in wild broodstock fish guts. (B) The Pluerocapsa clade example is more abundant in traditional hatchery broodstock fish. **Figure 6.** Examples of gut bacterial clades that are more abundant in either traditional hatchery or wild broodstock fish. Asterisk (\*) indicates a statistically significant result using Kruskal-Wallis tests and false discovery rate multiple test correction. (**A**) This *Peptonophilus* clade example is more abundant in wild broodstock fish guts. (**B**) The *Pluerocapsa* clade example is more abundant in traditional hatchery broodstock fish. Figure 6. Examples of gut bacterial clades that are more abundant in either traditional hatchery or wild broodstock fish. Asterisk (\*) indicates a statistically significant result using Kruskal-Wallis tests and false discovery rate multiple test correction. (A) This Peptonophilus clade example is more abundant in wild broodstock fish guts. (B) The Pluerocapsa clade example is more abundant in traditional hatchery broodstock fish.

Figure 7. Examples of gut bacterial clades that are more abundant in all Cedar Creek Hatchery samples compared to all East Fork Alsea Hatchery samples. Asterisk (\*) indicates a statistically significant result using Kruskal-Wallis tests and false discovery rate multiple test correction. (A) A clade from the genus Flavobacterium that is more abundant in Cedar Creek Hatchery fish guts. Different Bacteriodes clades that are (B) more abundant in East Fork Alsea Hatchery fish guts and (C) more abundant in Cedar Creek Hatchery fish guts. Figure 7. Examples of gut bacterial clades that are more abundant in all Cedar Creek Hatchery samples compared to all East Fork Alsea Hatchery samples. Asterisk (\*) indicates a statistically significant result using Kruskal-Wallis tests and false discovery rate multiple test correction. (A) A clade from the genus Flavobacterium that is more abundant in Cedar Creek Hatchery fish guts. Different Bacteriodes clades that are (B) more abundant in East Fork Alsea Hatchery fish guts and (C) more abundant in Cedar Creek Hatchery fish guts. **Figure 7.** Examples of gut bacterial clades that are more abundant in all Cedar Creek Hatchery samples compared to all East Fork Alsea Hatchery samples. Asterisk (\*) indicates a statistically significant result using Kruskal-Wallis tests and false discovery rate multiple test correction. (**A**) A clade from the genus *Flavobacterium* that is more abundant in Cedar Creek Hatchery fish guts. Different *Bacteriodes* clades that are (**B**) more abundant in East Fork Alsea Hatchery fish guts and (**C**) more abundant in Cedar Creek Hatchery fish guts.

Figure 8. NMDS plots and PAM cluster analysis reveal two clusters that may represent two steelhead gut microbiome types. (A) shows the combined NMDS visualization of all wild-born and hatchery-reared steelhead gut microbiome samples. The samples aggregate into two groups separated by space in the ordination. The coloring based on river basins and the shapes based on management type indicate that neither of these variables separates out into these two clusters. (B) colors each microbiome sample based on the PAM cluster designation and shows that PAM cluster assignment corresponds with the two speculated clusters. **Figure 8.** NMDS plots and PAM cluster analysis reveal two clusters that may represent two steelhead gut microbiome types. (**A**) shows the combined NMDS visualization of all wild-born and hatcheryreared steelhead gut microbiome samples. The samples aggregate into two groups separated by space in the ordination. The coloring based on river basins and the shapes based on management type indicate that neither of these variables separates out into these two clusters. (**B**) colors each microbiome sample based on the PAM cluster designation and shows that PAM cluster assignment corresponds with the two speculated clusters.

#### 4. Discussion **4. Discussion**

Pacific salmonid fisheries have the task of keeping up with consumer demands as wild and hatchery population numbers decline. Understanding how the Pacific salmonid gut microbiome varies based on broodstock ancestry or geographic location will provide insight into how gut bacteria may be manipulated to improve fish health and survival. This study defines how the steelhead trout gut microbiome varies across three river basins and as a function of their broodstock background. In particular, this study reveals geographic, geographic by phylogenetic lineage, and ancestry effects on the steelhead gut microbiome. Additionally, this study found an association between the steelhead gut microbial community and weight, which may have fitness implications for these fish. We document several bacterial clades that stratify groups with differing gut microbial diversity. Finally, a combined analysis revealed two predominant types of steelhead gut microbiome composition. This work clarifies how geographic location and broodstock affect the steelhead gut microbiome and informs our understanding of how the gut microbiome manifests in declining fish populations, which may lead to improved management practices or conservation efforts. This study highlights the existence of geographic effects that influence the composi-Pacific salmonid fisheries have the task of keeping up with consumer demands as wild and hatchery population numbers decline. Understanding how the Pacific salmonid gut microbiome varies based on broodstock ancestry or geographic location will provide insight into how gut bacteria may be manipulated to improve fish health and survival. This study defines how the steelhead trout gut microbiome varies across three river basins and as a function of their broodstock background. In particular, this study reveals geographic, geographic by phylogenetic lineage, and ancestry effects on the steelhead gut microbiome. Additionally, this study found an association between the steelhead gut microbial community and weight, which may have fitness implications for these fish. We document several bacterial clades that stratify groups with differing gut microbial diversity. Finally, a combined analysis revealed two predominant types of steelhead gut microbiome composition. This work clarifies how geographic location and broodstock affect the steelhead gut microbiome and informs our understanding of how the gut microbiome manifests in declining fish populations, which may lead to improved management practices or conservation efforts.

tion of the gut microbiome. These observations generally agree with prior studies of wildlife gut microbiomes in terrestrial systems [21,22], and a recent meta-analysis revealed differences in the gut microbiome of over 85 species of fish based on the five Korean water sources they were sampled from [23]. Another study, though, found that the wild gut microbiome of Atlantic salmon did not associate with geographic location [24]. However, our study was conducted with a larger sample size of Pacific salmonids, suggesting a This study highlights the existence of geographic effects that influence the composition of the gut microbiome. These observations generally agree with prior studies of wildlife gut microbiomes in terrestrial systems [21,22], and a recent meta-analysis revealed differences in the gut microbiome of over 85 species of fish based on the five Korean water sources they were sampled from [23]. Another study, though, found that the wild gut microbiome of Atlantic salmon did not associate with geographic location [24]. However, our study was conducted with a larger sample size of Pacific salmonids, suggesting a larger effect size many be needed to reveal geographic patterns in salmonids or that differences between Atlantic and Pacific salmonids—such as differences in physiology, ecology, or geography may account for these distinct results.

Importantly, cryptic variation in host physiology or genetic background may shape the gut microbial composition in this study observing wild-born gut microbes. Salmonids show evidence of subpopulations and genetic differences even within the same river system, and genetic differences have been seen in trout with spawning habitats as low as 2 km apart [25,26]. Given that the host genotype plays a role in shaping gut microbial composition in other fish hosts, the differences seen in the gut microbiome across geographic locations may be related to the accompanying differences in host genetics [27]. However, the steelhead genetics of the wild samples were not explicitly documented in this study, and future work should attempt to correlate gut microbial members with genetic differences in wild steelhead trout.

We uncovered that specific bacterial clades were more abundant in one of the three river basins, which supports the hypothesis that gut microbes may be useful for assessing salmonid biogeography. For example, a clade belonging to the genera *Ferruginibacter* was more abundant in the Alsea Basin, a Sphingomondaceae clade was more abundant in the Nestucca Basin, and *Novosphingobium* was more abundant in the Siletz Basin. In a previous study, *Novosphingobium* abundance varied by geographic location in the gut of another fish species, suggesting that members of this bacterial genera typically show geographic patterns within the fish gut microbiome [28]. While the function of *Ferruginibacter* in the fish gut is unknown, bacteria from this genus are often isolated from freshwater sediment, suggesting that these bacteria are dispersing from sediment to fish gut or from fish gut to sediment [29]. Regardless, given the cross-sectional nature of our study, it is not clear if these geographic associations are maintained over the course of a fish's lifespan, a topic that should be explored in future work.

An additional analysis revealed several clades that display correlations between geographic distance and phylogenetic distance in the steelhead gut microbiome, suggesting that these gut bacterial members co-diversified with their hosts. Alternatively, these bacterial clades may manifest a geographic distribution in the environment and then occupy the fish host. Clades demonstrating this geographic and phylogenetic correlation include clades from the bacterial families Rhodobacteraceae and Sphingomonodaceae. Rhodobacteraceae may play a role in fish health, as this family was previously found to be more abundant in the guts of healthy shrimp compared to diseased shrimp [30], but future studies should explicitly test its role in steelhead health. Also, bacterial clades from Sphingomonodaceae produce sphingolipids, which are organic compounds that can modulate *O. mykiss* mucosal homeostasis and B cell abundance [31]. Although we did not sample mucosal-associated bacteria, mucosal membranes and digesta share some microbial members and microbes in the digesta and lumen can still produce compounds that affect host immune responses [32]. Additionally, Sphingomonadaceae possess sphingolipids in their cell membranes that improve chances of successful colonization and survival in the gut, which can be advantageous for both commensal and pathogenic organisms [33,34]. This speculative role of Rhodobacteraceae and Sphingomonadaceae may be the reason for a potentially prolonged association between these bacterial families and steelhead that gave rise to this geographic lineage sorting.

Despite differences in gut microbial structure across basins, we discovered bacterial clades that are prevalent in all steelhead guts of our first study. Bacterial clades from the genera *Flavobacterium*, *Hyphomicrobium*, and *Singulisphaera* represent such core taxa. The ubiquitous presence of these bacteria suggest steelhead physiology selects for these specific clades, as they may have critical functions within the steelhead gut, or that these bacterial clades are also commonly found in the surrounding aquatic meta-communities. The function of *Hyphomicrobium* and *Singulishpaera* in the gut are unknown, but they have been found in aquatic systems as well as other fish guts [35,36]. Several members of the *Flavobacterium* genus are pathogenic to fish, although some *Flavobacterium* are commensal [37]. The pathogenicity of the *Flavobacterium* clades in this study is unclear but could have widespread consequences, as these clades were found in every fish of our first study. Given

their ubiquitous distribution, future investigations should seek to discern the physiological impacts of these taxa.

In addition, this study clarifies the impact of hatchery broodstock on the gut microbiome. Previous work suggested that hatcheries elicit strong selective pressure on Pacific salmonids that differentiate fish reared in hatcheries for several generations from fish reared in hatcheries for one generation, who are both different from wild-born salmonids. For example, the relative fitness levels and rates of reproductive success of fish with greater hatchery ancestry are significantly lower than those of fish with wild ancestry [38]. Also, the expression of several genes from the first generation of hatchery steelhead trout (i.e., previously wild trout) are heritably altered after a single generation in a hatchery environment [39]. The differences in the diversity of traditional hatchery and first-generation hatchery gut microbiomes suggest this selective pressure is also applied to steelhead gut microbial communities. Other heavily managed animals are known to have different gut microbial communities compared to their wild counterparts, as is the case with animals in captivity [40].

The fact that the first-generation hatchery stock and traditional hatchery stock were reared in the same hatchery environment suggests that the differences in the gut microbiome between these two groups is due to differences in genetics. Genotype has previously affected gut microbial composition in fish and other hosts [27,41]. Future conservation efforts may use the identification of specific clades that stratify or are indicative of a hatchery or wild steelhead gut microbiome to identify a fish as early generation or traditional hatchery-reared. Additionally, our resolution of clades that differentiate traditional hatchery and first-generation hatchery fish microbiomes may help hatcheries develop management practices that ultimately normalize the composition of hatchery-reared microbiomes closer to their wild counterparts. This study only focused on the gut microbiome of juvenile steelhead as this is the life stage steelhead are contained in hatcheries, and more mature steelhead undergo a great deal of physiological changes in preparation to travel out to the ocean, which may induce changes in the gut microbiome.

A combined analysis using all steelhead gut microbiome samples from our two studies revealed two robust clusters that demonstrate densely populated areas in the multidimensional space of steelhead gut microbiome beta diversity. These two clusters may be evidence of two different steelhead gut microbiome types. However, an unknown covariate that we did not measure, such as the sex of the fish, may be responsible for the clustering. Future studies may find these two clusters are robust among other populations of steelhead gut microbiomes and they should focus on measuring more variables that may be causing these clusters. If these two clusters are robust, future researchers should consider that the effectiveness of microbial interventions may be different based on the steelhead microbiome type measured in future studies. Therefore, microbiome type should be another variable considered when studying the steelhead gut microbiome.

Our results indicate that the steelhead gut microbiome varies as a function of geography and broodstock ancestry. Additionally, several steelhead gut bacterial clades show geographic lineage sorting across western Oregon, and a collective analysis showed two gut microbiome types. Given the declining populations of wild salmonids and the comparatively poor fitness of successive generations of hatchery-reared supplementation stock, characterizing the gut microbial communities across these populations is critical in learning 1) how the steelhead gut microbiome plays a role in the health and fitness of these fish and 2) how we can use steelhead gut microbiota or microbial interventions to improve conservation and supplementation efforts.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms10050933/s1, Table S1: Number of reads before and after rarefaction for the geography study; Table S2: Table of clades that were correlated with phylogenetic distance and geographic distance. The table includes mantel test *p*-values and associated bacterial taxa identification; Table S3: Number of reads before and after rarefaction for the broodstock study; Table S4: Gut microbial clades significantly different across traditional hatchery broodstock

and wild broodstock fish; Table S5: Gut microbial clades significantly different between North Fork Alsea Hatchery and Cedar Creek Hatchery.

**Author Contributions:** Conceptualization, N.S.K., S.C. and T.J.S.; data curation, N.S.K., T.C. and S.S.; methodology, N.S.K., S.C. and T.J.S.; resources, T.C., S.S., S.C. and T.J.S.; formal analysis, N.S.K. and T.J.S.; writing—original draft preparation, N.S.K. and T.J.S.; writing—review and editing, N.S.K., T.C. and T.J.S.; visualization, N.S.K.; supervision, T.J.S.; funding acquisition, N.S.K. and T.J.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by The National Science Foundation (grant 1557192), institutional funds to T.J.S., and an Oregon Department of Fish and Wildlife fellowship to N.S.K.

**Institutional Review Board Statement:** Animal handling, sampling, and necropsy were conducted by the Oregon Department of Fish and Wildlife as part of their routine management and conservation efforts in accordance with state permits. Microbiome samples were collected from animals that were otherwise subject to these efforts; no animals were handled, sampled, or sacrificed specifically for this study.

**Data Availability Statement:** DNA and R code generated during this research is located on the Sharpton Lab Repository (http://files.cqls.oregonstate.edu/Sharpton\_Lab/Papers/Kirchoff\_Microorganisms\_ 2022/ accessed on 22 April 2022).

**Acknowledgments:** We would like to thank Mark Dasenko at the Center for Quantitative Life Sciences for sequencing samples and providing advice on sample preparation. We thank Jerri Bartholomew for helpful discussions. We would also like to thank the employees at North Fork Alsea Hatchery and Cedar Creek Hatchery for allowing us to sample fish and Jamie Anthony for providing additional historical information about the two hatcheries.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

