Genotypic Diversity and Host-Specificity of Frankia Bacteria Associated with Sympatric Populations of Alnus rubra and Alnus rhombifolia in Oregon

Abstract Biological nitrogen fixation is one of the most critical processes contributing to ecosystem productivity and stability on a global scale. In temperate climates of the northern hemisphere, plant-root associated bacteria of the genus Frankia are the major nitrogen fixers in forest environments. Trees belonging to the genus Alnus are the most widespread hosts of Frankia in the Pacific Northwest, and a myriad of biotic and abiotic factors can influence the robustness of this symbiosis. Host identity and bacterial strain are important features that can impact Alnus-Frankia association, but little is known about the interplay of intrageneric hosts that co-occur in natural settings. In this study we investigated the genetic diversity and host specificity of Frankia bacteria associated with sympatrically occurring populations of Alnus rubra (red alder) and Alnus rhombifolia (white alder) in Oregon. Based on sequence analysis of the nifH gene recovered from root nodules we found low overall bacterial diversity. One dominant Frankia genotype was associated with both host species, indicating a lack of strong host specificity in this system. Our results suggest that certain intrageneric plant hosts with overlapping distributions show cross-compatibility with symbiotic actinorhizal bacteria, and that low strain diversity of these bacteria can persist across mixed host populations.


Introduction
Nitrogen-fixing bacteria associated with plant roots are one of the primary sources of biological nitrogen inputs in terrestrial ecosystems. In temperate climates, bacteria of the genus Frankia are the major group that forms such symbioses with 24 actinorhizal angiosperm genera worldwide (Schwintzer andTjepkema 1990, Benson andSilvester 1993). Members of the genus Alnus (alders) are the only known trees that form associations with Frankia bacteria, and Alnus rubra Bong. (red alder) is one of the most common tree species in the northwest (Franklin and Dyrness 1973). Accordingly, A. rubra represents a critical host for actinorhizal bacteria in the Pacific Northwest owing to its widespread distribution and relatively long lifespan compared to non-tree actinorhizal hosts in this region. Alders are pioneer tree species that can not only enhance soil nitrogen content, but can also lower soil pH (Lawrence 1958), significantly increase soil organic matter (Tarrant and Miller 1963, Franklin et al. 1968, Bormann and DeBell 1981, Binkley et al. 1982, Binkley 1983, and improve the retention of important anions such as PO 4 3and SO 4 2- (Johnson et al. 1986). Furthermore, the soil enhancing capacity of Frankia-associated Alnus species can also benefit neighboring trees of economic importance such as Pseudotsuga menziesii (Mirb.) Franco (Douglasfir) (Miller and Murray 1978) and Pinus species (Dai et al. 2004), with which they are sometimes interplanted for this purpose.
The performance of the actinorhizal symbiosis can vary significantly depending on host plantbacterial genotype combinations (Wheeler et al. 1986, Weber et al. 1987. While a number of different factors can influence the genotypic diversity of Frankia including edaphic (Sheppard et al. 1989, Crannell et al. 1994, Nickel et al. 1999) and biogeographic (Khan et al. 2007, Põlme et al. 2014 factors, the effect of host identity is of particular interest and has been widely studied (van Dijk et Diversity of Frankia in Alnus al. 1988, Rouvier et al. 1996, Navarro et al. 1999, Vanden Heuvel et al. 2004, Huguet et al. 2004, Lipus and Kennedy 2011, Põlme et al. 2014. Four major host infection groups have been identified for Frankia based on their ability to form nodules within actinorhizal plant families (Baker 1987), though it has since become clear that strains do not always remain within the boundaries of this host-specific classification (Torrey 1990). Much of the work examining Frankia host specificity in Alnus has focused on strain compatibility across allopatric populations of different (Anderson et al. 2009, Lipus and Kennedy 2011, Põlme et al. 2014 or the same ) host species. Few studies have investigated genotypic diversity and host specificity of the bacterial symbiont in sympatric populations of different Alnus species (Anderson et al. 2009, Pokharel et al. 2011, beckoning the questions of how proximity, shared soil environment, and exposure to the same Frankia strain pool act to influence host colonization and specificity. Pokharel et al. (2011) conducted a comprehensive assessment of Frankia associated with 12 Alnus taxa-most of which do not typically co-occur in their natural habitats-growing sympatrically in an arboretum setting. Using Rep-PCR (repetitive element polymerase chain reaction) fingerprinting, Pokharel and colleagues (2011) generated genetic strain profiles for nodules collected from multiple individuals of each taxon. Strikingly, only three distinct strain groupings were observed, with Frankia associated with nine of the Alnus taxa comprising a single strain (I), associates of two other taxa comprising a second strain (II), and a single taxon associate comprising the third group (III), suggesting low genetic diversity and high intrageneric host cross-compatibility. In contrast, Anderson et al. (2009) observed a total of nine unique RFLP (restriction fragment length polymorphism) profiles generated from Frankia nodules from A. viridis and A. tenuifolia populations growing sympatrically in interior Alaska, three of which occurred on both hosts and three of each of the remaining six profiles occurring on either host. Interestingly, the most abundant genotypes in each of three sites were consistently those that were unique to each host suggesting a greater degree of host specificity for those profiles.
Given such mixed results, the paucity of more accurate and replicable molecular approaches (e.g., sequence-based genotyping), and the underrepresentation of some Alnus species in previous studies, there is a need to further investigate the diversity and host specificity of actinobacteria among sympatrically distributed populations of alders in natural settings. Of the four Alnus species that are native to the Pacific Northwest (including northern California), red alder (Alnus rubra) and white alder (Alnus rhombifolia Nutt.) are the most common (Little 1976) and have overlap in their distribution. Here we assessed the genetic diversity and host specificity of Frankia bacteria associated with the roots of sympatric populations of A. rubra and A. rhombifolia in western Oregon.

Study Area
Nodule-bearing roots were collected from both red and white alder from a site in Sherwood, western Oregon (Heaton Creek, Figure 1). Alder species were identified considering a set of traits including variation in leaf shape, flowers, and extent and speed of oxidation processes resulting in red coloration of scratched inner tree bark. Red alder periderm (in line with its common name) shows a faster and much more intense coloration in response to physical damage compared to white alder. In total, 28 white alder and 30 red alder samples were collected, with each sample originating from a different tree. Trees had diameters at breast height between 10 and 25 cm. Both species of trees showed similar variation in diameter (red alder: 16.4 ± 4.0, white alder: 15.9 ± 4.7 [mean ± SD]). Nodule bearing roots were collected from the upper soil level (15 cm) at a radius of 50 cm around each tree. Nodules were only collected from larger roots (> 1 cm in diameter) easily trackable to the specific sample tree or from finer roots that were clearly attached to these larger roots. After collection, the roots with nodules were stored in sealed zip lock bags and transported to the lab in a cooler (4 °C). Nodules were processed within 48 hours.

Surface Sterilization
Roots and attached nodules were washed with deionized water to remove soil and organic matter. Nodules were then removed from roots with a sterile scalpel and placed into individual 14 mL sterile scintillation vials. Each vial was then filled with 10 mL of 10% bleach (0.6% hypochlorite) solution and agitated in an orbital shaker at room temperature (21 °C) for 2 minutes at 140 rpm. After agitation, the bleach solution was decanted from each vial and the bleach washing process was repeated once more. After the second bleach wash, the washing process was repeated three more times with deionized water instead of bleach. After surface sterilization, a single nodule lobe of each sample was removed with a sterile scalpel. All single lobes were then transferred to 1.5 mL microcentrifuge tubes for DNA extraction using a sterile toothpick.

DNA Extraction
DNA was extracted from individual nodule lobes using the Sigma Tissue Extract-N-Amp Kit (Sigma-Aldrich, St. Louis, MO). Forty μL of extraction buffer were added into each 1.5 mL microcentrifuge tube containing an individual nodule lobe, which was then homogenized using a sterile micropestle. Lobe homogenates were centrifuged for 1 minute at 15,000 rpm (21,130 x g), and 20 μL of supernatant from each tube transferred into an individual 0.2 mL PCR strip-tube. Strip-tubes containing extractions were placed in a thermal cycler and incubated at 65 °C for 10 minutes, followed by 95 °C for 10 minutes. Following incubation, 25 μL of neutralization buffer were pipetted into each sample which were then briefly vortexed and stored at 4 °C until use.

PCR and Sequencing
A 606 bp (base pair) portion of the nifH gene was amplified by PCR using Frankia-specific primers nifHf1 (5′-GGC AAG TCC ACC ACC CAG C-3′) and nifHr (5′-CTC GAT GAC CGT CAT CCG GC-3′). PCR reactions were set up in 24 μL volumes containing 8.45 μL PCR water, 12.5 μL GoTaq Master Mix, 1.25 μL BSA (1 mg/1 mL), 0.4 μL of each primer (10 μM), and 1 μL of 1:10 diluted template. Reaction mixtures were then subjected to the following thermal cycling conditions: 96 °C for 5 minutes, 35 cycles at 95 °C for 30 seconds, 60 °C for 30 seconds, 72 °C for 45 seconds, and   (Normand et al. 2007) was downloaded from GenBank and included in the alignment as an outgroup. The ends of the alignments were trimmed to ensure each sequence was of the same length.

Phylogenetic Analysis and Statistics
To determine the optimal nucleotide substitution model for use in phylogenetic analysis, the alignment of OTUs created from both Alnus spp. at the 97% similarity threshold was analyzed using PartitionFinder version 2.1.1 (Lanfear et al. 2016).
The entire alignment was treated as a single data block, "models" and "schemes" were set to all, and model selection was done based on the corrected Akaike Information Criterion. This analysis indicated that a general time reversible model with an estimation of invariant sites (GTR+I) was the best fit, and this model was used in both maximum likelihood and Bayesian tree building.
Maximum likelihood trees were built in Geneious using RAxML version 8.2.11 (Stamatakis 2014) with the "GTR CAT I" model setting and the Rapid Bootstrapping algorithm; 10,000 bootstrap replicates were performed, and the consensus trees were built with a 25% bootstrap support minimum. Bayesian trees were built using BEAST version 1.8.4  with the GTR+I nucleotide model. A strict clock model was used, along with a coalescent constant-population tree prior. The program was run with chain lengths of 10,000,000 states, with sampling every 1,000 states. Maximum clade credibility trees were generated in Tree Annotator with a 10% burn-in.
The software EstimateS 9.1.0 (Colwell 2013) was used to estimate the true richness of Frankia genotypes associated with A. rubra and A. rhombifolia. The Chao2 estimator was computed on the basis of 500 randomizations of sample order without replacement. To assess whether the host identity significantly influenced the co-occurrence of the two most frequently observed Frankia genotypes, Fisher's Exact Probability Test was performed.

nifH Sequence Analysis
We sequenced a portion of the nifH gene from Frankia nodules associated with the roots of 28 Alnus rhombifolia and 30 Alnus rubra individuals in order to assess the genetic diversity and level of host specificity of these actinobacteria in sympatric populations of their hosts. Of the 58 nodules processed we successfully generated nifH sequences for 47, which were comprised of five OTUs (herein referred to as genotypes) at 97% sequence similarity across both tree species. Sequences derived from both A. rubra and A. rhombifolia associated nodules clustered into each of five genotypes at varying proportions with A. rubra derived sequences outnumbering A. rhombifolia derived sequences in four out of the five genotype groupings (Table 1). The calculated Chao2 richness estimator was identical to the observed richness of five Frankia genotypes shared by the two Alnus populations, illustrating that sampling of nodules was adequate to represent true genotypic richness. More than 50% of sequences derived from both Alnus host species clustered into one dominant genotype (DJB-F1), which was associated with trees that were heterogeneously distributed across the site (Figure 1). Similarly, trees associated with the other four genotypes (DJB-F2-DJB-F5) also exhibited heterogeneous spatial distribution. Upon comparing the Alnus-associated Frankia genotypes in our study to those characterized in other studies conducted in our region (Kennedy et al. 2010, Lipus andKennedy 2011), we discovered that two of our genotypes (DJB-F1 and DJB-F5) were identical to two described previously (KL2 and KL1.6 respectively). The results of the Fisher's Exact Probability Test (P = 0.618) indicate that no systematic association exists between Alnus species and occurrence of Frankia genotype in at this site.

Phylogenetic Analysis
Both maximum likelihood and Bayesian phylogenetic analyses of the five Frankia genotypes recovered from A. rubra and A. rhombifolia yielded trees with identical topologies ( Figure  2). The genotypes are clustered into two major clades, with one clade consisting of DJB-F3, and the other consisting of two monophyletic subclades containing two of each of the remaining four genotypes (DJB-F1, DJB-F5; and DJB-F2, DJB-F4, respectively). Relatively high bootstrap support values and posterior probabilities support the stability of these groupings.

Discussion
The goal of the current study was to assess genetic diversity and host specificity of actinorhizal Frankia bacteria associated with the roots of two co-occurring Alnus taxa. Our findings suggest that host identity is not a strong driver of Frankia genotype abundance associated with sympatric populations of Alnus rubra and A. rhombifolia. Whether or not these findings are applicable on a larger scale remains elusive as we were limited to a single location in western Oregon. Still, to the best of our knowledge, this is the first study comparing host-Frankia associations of these species at a natural site.
Similar to other studies we found low overall diversity of Frankia (Benson et al. 1996, Clawson et al. 1997, Oakley et al. 2004, Vanden Heuvel et al. 2004, Lipus and Kennedy 2011, Pokharel et al. 2011) with a single dominant genotype across the sampling site. Interestingly, two of the genotypes recovered in the current study were identical to two described in previous studies ( Table 1) that were found at sites between 50 km and 115 km from the current study site. Remarkably these matching genotypes correspond to the most abundant genotype in our study and the two most abundant in those of Kennedy et al. (2010) and Lipus and Kennedy (2011), further underscoring the regional low diversity of Alnus-associated Frankia. There were no genotypes that were unique to either host, and no significant association between the abundance of genotypes and host species was observed indicating a lack of host specificity in this system. A similar pattern was observed by Pokharel et al. (2011) who found that Rep-PCR profiles (i.e. genotypes) of Frankia associated with nine out of 12 Alnus taxa in a common garden setting were identical. Likewise, Oakley et al. (2004) performed a sequence-based analysis of Frankia associated with several co-occurring Ceanothus species and reported a high level of sequence homogeneity across hosts despite not having clustered similar sequences into OTUs, which would have likely further reduced the observed diversity. Interest-  , Lipus and Kennedy 2011, Põlme et al. 2014. In contrast, the findings of Anderson et al. (2009) revealed that the most abundant Frankia strains (based on RFLP profiles) recovered from overlapping A. viridis and A. tenuifolia were unique to each host, in that case indicating a high degree of host specificity in their study site.
Whereas strict host specificity appears to be more the exception than the rule within intrageneric hosts (especially those that are sympatrically distributed), variation in this phenomenon is likely to be influenced by host phylogeny and not simply taxonomic identity. This was more recently highlighted in a large-scale global assessment of Frankia associated with 22 Alnus species on four continents by Põlme et al. (2014). From their investigation, Põlme and associates discovered that host phylogeny was the primary driver of Frankia assemblages worldwide and that the majority of recovered genotypes were generalists, with a few that were strongly host specific. This may explain the mixed results obtained across other studies comparing the Frankia assemblages of co-occurring hosts of the same genus, whereby different host species combinations may produce different outcomes in respect to symbiont affinity. The Alnus species in our study are more closely related compared to other similar studies (Chen and Li 2004) in which a high degree of host preference (Lipus and Kennedy 2011) or host specificity (Anderson et al. 2009) were found, supporting the idea that host phylogeny may reflect the degree of homogeneity in Frankia populations associated with co-occuring host species. It is also noteworthy that Põlme et al. (2014) found that Frankia phylogeny did not significantly influence the identity of associated hosts, reinforcing the idea that the Alnus host is selecting the symbiont and not vice versa. This may also explain the oddity of the dominant genotype in our study being most closely related to the rarest in our study (Figure 2). In other words as suggested above, perhaps selective initiation of the actinorhizal mutualism depends on factors not bound to symbiont phylogeny, or at least not to the degree that is revealed by variation in the nifH gene.
While certain key features such as low overall diversity with few dominant genotypes appear to be consistent across studies, the reason for this phenomenon is still elusive. One possible explanation for the low diversity of host-associated Frankia are potential limitations in resolution of using single standard marker genes like nifH. However, the nifH gene currently represents by far the most utilized marker for studying genotypic variation in nitrogen-fixing bacteria-including Frankia-and many key studies on Frankia diversity available today have used this marker , Mirza et al. 2009, Kennedy et al. 2010, Põlme et al. 2014, Rodriguez et al. 2016. Other new markers (or combinations thereof) such as the pgk region for example (Pozzi et al. 2018) are promising to provide greater analytical depth but are little tested so far or in the case of multiple markers often yield conflicting results. While nifH has limitations it still is useful to provide data with the particlar advantage of being easily comparable to existing studies databases and allowing the specific targeting of Frankia bacteria in environmental samples. Ultimately, full genomic comparisons will eventually yield the best metric for deriving genotypic grouping of Frankia strains.
Another potential explanantion for the relatively low diversity of host-associated Frankia is that only few Frankia genotypes provide significant adaptive advantages to host plants in nature. At this point, the question why and under which conditions host associations with few dominant Frankia genotypes-as opposed to diverse consortia-provide an adaptive advantage to host fitness remains unsanswered. Bioassays with specific host-Frankia associations under controlled conditions would provide useful tools to study host-Frankia specificity (Lipus and Kennedy 2011), but whether or not different symbioses provide actual advantages or disadvantages in nature requires bioassays at natural sites. Thus, future efforts would benefit from the inclusion of multiple combinations of co-occuring host species, and detailed studies on the prevalence of specific host-Frankia combinations under variable micro-climatic conditions in nature. Combining such outdoor studies with measuring phenotypic responses of host plants in combinatorial experiments under controlled conditions will help illuminate the functional underpinnings of these unique associations.

Acknowledgments
Funding by the National Science Foundation (NSF) to DJB (grants IOS 1457369 and 1656057) is gratefully acknowledged. The authors would like to thank Christopher Diego for assistance with site map design, and gratefully acknowledge funding by the National Science Foundation (NSF) to DJB (grants IOS 1457369 and 1656057).