Are migratory waterfowl vectors of seagrass pathogens?

Abstract Migratory waterfowl vector plant seeds and other tissues, but little attention has focused on the potential of avian vectoring of plant pathogens. Extensive meadows of eelgrass (Zostera marina) in southwest Alaska support hundreds of thousands of waterfowl during fall migration and may be susceptible to plant pathogens. We recovered DNA of organisms pathogenic to eelgrass from environmental samples and in the cloacal contents of eight of nine waterfowl species that annually migrate along the Pacific coast of North America and Asia. Coupled with a signal of asymmetrical gene flow of eelgrass running counter to that expected from oceanic and coastal currents between Large Marine Ecosystems, this evidence suggests waterfowl are vectors of eelgrass pathogens.

strain of protist, Labyrinthula zosterae. This outbreak resulted in declines in eelgrass-dependent fish and waterfowl species (Short, Muehlstein, & Porter, 1987), including a collapse of the Atlantic Flyway population of Atlantic Brant, B. b. hrota (Cottam, Lynch, & Nelson, 1944;Kirby & Obrecht, 1982); however, L. zosterae is not the only pathogenic organism known to impact eelgrass health. In 2016, Govers et al. (2016) reported the presence of two other closely related fungi-like oomycete species associated with eelgrass in the Atlantic: Phytophthora gemini and a previously undescribed species, Halophytophthora sp. Zostera. Both oomycete species are potent pathogens closely related to the potato blight (P. infestans), and both may strongly reduce sexual reproduction in eelgrass by reducing seed germination (Govers et al., 2016). Diseases specific to eelgrass continue to play an important role in regulating eelgrass populations in Europe and the Atlantic coast of North America (Bishop, Martin, & Ross, 2017;Govers et al., 2016;Muehlstein, 1989;de los Santos et al., 2019;Short et al., 1987) and may play a future role along the Pacific coast (Martin et al., 2016).
Current research indicates eelgrass pathogens may be transmitted via direct contact with infected tissue and/or debris (Martin et al., 2016;Muehlstein, 1992) or via waterborne transmission of other infected substrates (Martin et al., 2016). It is unclear how these eelgrass pathogens are vectored in the north Pacific, although certainly oceanographic and coastal currents play a role in eelgrass dispersal (Kendrick et al., 2012), and so presumably in the dispersal of associated pathogens. Nevertheless, a signal of counter-current gene flow between certain eelgrass meadows along the Alaska Peninsula suggests oceanic and coastal currents alone are not the only dispersal mechanism for eelgrass populations in the region (Talbot et al., 2016). Further, there is an unexpectedly close genetic relationship between eelgrass meadows in Kinzarof and Izembek lagoons (Talbot et al., 2016), two lagoons separated by only 5 km of land, but at the least 510 km of coastline via the Alaska Coastal Current (ACC) that flows northward from the Gulf of Alaska Large Marine Ecosystem (GoA-LME) into the Eastern Bering Sea Large Marine Ecosystem (EBS-LME) at Unimak Pass ( Figure 1). This finding suggests an additional dispersal mechanism-in particular, waterfowl (Talbot et al., 2016)-for eelgrass, and thus for pathogens on eelgrass.
To investigate potential links between migratory bird species, eelgrass communities, and vectors of eelgrass pathogens along the F I G U R E 1 Environmental DNA sampling locations. Environmental DNA sample locations (in green) and prevailing direction of the Alaska Coastal Current (ACC) in the Gulf of Alaska Large Marine Ecosystem (GoA-LME) and Eastern Bering Sea Large Marine Ecosystem (EBS-LME). IZL, Izembek Lagoon; KIL, Kinzarof Lagoon. Green indicates extent of eelgrass meadows in IZL and KIL Alaska Peninsula, we leveraged environmental DNA metabarcoding to (a) detect the presence of DNA of both classes of pathogens in environmental samples (sediment, water column, plant), (b) test for seasonal differences in their presence, and (c) determine whether DNA from eelgrass pathogens ( Figure 2) occurred in the cloaca of waterfowl species that forage on eelgrass. The presence of eelgrass pathogens in cloacal contents of waterfowl would suggest the potential for vectoring of the pathogens among embayments via endochory-that is, when plants are eaten and dispersed by animals. To test this hypothesis, we analyzed gene flow (Beerli & Felsenstein, 1999, 2001  F I G U R E 2 Labyrinthula spp. phylogeny. Labyrinthula spp. phylogeny, based on partial 18S rDNA sequences, reconstructed from Martin et al. (2016) and replicating their analyses. Maximum-likelihood (1,000×) and neighbor joining (500×) bootstrap values are shown in the nodes, respectively. Red circles in the node identify the pathogenic clade, and OTUs, designated by GenBank Accession numbers corresponding to nucleotide sequence; highlights in red indicate strains known to be pathogenic based on Martin et al. (2016). OTUs designated by asterisks represent sequences recovered from environmental samples collected from Grant Point, Izembek Lagoon, AK

| Sample collection
Environmental samples were collected from the water column, sediment, and eelgrass leaves at Grant Point, Izembek Lagoon (55°16′12.40″N, 162°53′50.58″W), and from the cloaca of waterfowl species that had recently foraged in eelgrass meadows in Izembek Lagoon. Water samples were collected in 500 ml volumes and filtered through 0.22-micron filters (GTTP 04700, Millipore), which were then stored in 5 ml of Longmire Buffer (Longmire, Maltbie, & Baker, 1997) held in 15-ml falcon tubes (Menning, Simmons, & Talbot, 2018). Sediment samples were collected in 1 ml volumes and stored in 15-ml tubes containing 5 ml of LMB.
Approximately three inches of plant tissue (leaf) were collected and stored in 15-ml tubes containing 5 ml of LMB. Cloacal samples, which at times included seeds, were collected opportunistically from sport hunters that shot birds in the Cold Bay, AK, area dur-

| DNA extraction
Stored environmental samples were vortexed and eDNA extracted using a 400 µl subsample of the LMB-preserved sample using a Qiagen DNeasy Blood and Tissue kit (Qiagen), following manufacturers suggested protocols, with the exception that volumes were doubled. To avoid contamination, all extractions were conducted in a laboratory in which polymerase chain reactions (PCRs) have never been conducted and which is separated physically from laboratories where PCRs are conducted. Additionally, prior to this study, no studies involving seagrass pathogens were ever conducted in this laboratory, or any other laboratories located in the same facility.

| Primer and reference database design
Primers were designed using Python (Van Rossum, 1995) and Biopython (Cock et al., 2009) scripts that are part of the U.S.
Aligned FASTA files were used to locate potential primer sites (conserved regions greater than 17 base pairs). Once candidate primers were identified, they were checked against all locus-specific FASTA sequences on NCBI. The resultant FASTA files were screened to verify that a single taxon would be identified for each unique sequence and that no potential sequences had more than TA B L E 1 Taxon-specific pathogen primers designed for this study

| DNA library preparation and sequencing
Environmental DNA libraries were prepared using a two-step PCR protocol and sequenced using an Illumina MiSeq. PCR reac-

| Bioinformatic analyses
All demultiplexed data were retrieved from the Illumina MiSeq and analyzed in the same manner as  with the exception that the default BLAST+ parameters reward/penalty were changed to 1/−3, respectively, and the gapopen/gapextend parameters were set at 1/1 to ensure at least a 99% match to the reference database. Quality filtering to remove sequencing errors was conducted by including only match count information that exceeded 0.01% of the total number of reads passing filter, per sample, in the MiSeq run (Bokulich et al., 2013). Bioinformatic analyses were

| Assignment of pathogenicity of Labyrinthula
To assign any reads to the pathogenic versus nonpathogenic Labyrinthula clades, we reanalyzed sequence reads that were included in the phylogenetic tree comprising Figure 3 in Martin et al. (2016), which were identified to pathogenicity (see Figure 2). This facilitated our ability to estimate whether any Labyrinthula sequences recovered in our eDNA analysis could be assigned to virulent versus nonvirulent forms.

| Gene flow rates and polarity for eelgrass (Zostera marina)
We estimated the magnitude and polarity of gene flow among populations within the two regions using the maximum-likelihood approach implemented in MIGRATE 3.6.8 (Beerli & Felsenstein, 1999, 2001. MIGRATE uses a coalescence approach to estimate mutation- between population pairs. MIGRATE was performed using maximum-likelihood search parameters (10 short chains using 1,000 trees of 25,000 sampled followed by five long chains using 10,000 trees out of 250,000 sampled and five adaptively heated chains (start temperatures: 1, 1.5, 3, 6 and 12; swapping interval = 1). To ensure convergence of parameter estimates, full models were run ten times. We reported the results of one representational MIGRATE analysis in the text (also see Figure 4)

| Estimates of gene flow rates and polarity for eelgrass (Zostera marina)
Estimates of gene flow rates and polarity (Figure 4)

| Environmental DNA
This study reports the first instance in Alaska waters of pathogenic strains of Labyrinthula known to cause declines in seagrass meadows along the Atlantic coast of North America and Europe (Groner et al., 2014(Groner et al., , 2016 and two other seagrass pathogens, Phytophthora gemini and Halophytophthora sp. Zostera. Labyrinthula strains were detected in cloacal contents of eight of nine waterfowl species that annually migrate along the Pacific coast of North America and Asia (Derksen et al., 1996;Koehler, Pearce, Flint, Franson, & Ip, 2008;Lane & Miyabayashi, 1997). Although plant seeds can remain viable after passing through the waterfowl digestive system (Leeuwen et al., 2012), finding sequence reads from pathogenic taxa in waterfowl cloacal samples does not demonstrate that the organisms themselves remain viable.
Additional experiments are needed to demonstrate viability.

| Bioinformatic analyses
All demultiplexed Illumina MiSeq data can be found at NCBI BioProject PRJNA548352, and sample information can be found in

| D ISCUSS I ON
Izembek Lagoon is the first major intertidal eelgrass embayment waterfowl species encounter on the north side of the Alaska Peninsula during their southward migration from northerly breeding grounds. The lagoon contains the largest (16,000 ha) expanse of intertidal eelgrass on the Alaska Peninsula (Hogrefe, Ward, Donnelly, & Dau, 2014) and supports the greatest number of waterfowl in the region during fall migration (Wilson, 2019). During arrival, birds generally settle in Izembek Lagoon before moving on to Kinzarof Lagoon and other nearby eelgrass embayments on the peninsula (Boyd, Ward, Kraege, & Gerick, 2013). Therefore, F I G U R E 6 Zostera marina pathogens found in cloacal contents of migratory waterfowl. Percent of positive determinations of DNA from organisms with strains known to be pathogenic on eelgrass, found in cloacal contents of two species of waterfowl harvested by hunters in Cold Bay, Alaska, during September 2016, and nine species of waterfowl harvested during September 2018 the gene flow estimates are consistent with the hypothesis that eelgrass may be dispersed via endochory by waterfowl at levels sufficient to overcome the signal of gene flow facilitated by oceanic and coastal currents (Talbot et al., 2016), supporting a model for the transmission of associated pathogenic organisms via endochory. Thus, while gene flow and population dynamics for marine organisms, and presumably associated pathogens, are influenced by the distance between populations, currents, and oceanographic mixing patterns (Hedgecock, 1986), they may also be affected by movements of birds (Arasaki, 1950;Nacken & Reise, 2000;Sanchez et al., 2006;Sumoski & Orth, 2012).
The migratory waterfowl that stage on the lower Alaska Peninsula may consume large amounts of eelgrass (Lewis et al., 2013) and have large migratory ranges along the eastern and western sides of the Pacific (Derksen et al., 1996) that could have cascading ecological impacts on coastal marine communities. This intersection of multiple migratory ranges may facilitate the transmission of diseases outside of the range of any one particular migratory species. For example, Winker and Gibson (2010) found that migratory birds can move avian influenza from Asia to Alaska where they could infect other species and continue passing this infection across North America via other migratory pathways.
One of the earliest records of a marine conservation translocation involved eelgrass rhizomes translocated from the Pacific coast to the Atlantic of North America in 1943, following eelgrass decline in the Atlantic attributed to a pathogenic Labyrinthula infection (Cottam & Addy, 1947). In that instance, the successful transplants were destroyed by the disease the following year. While guidelines for translocation of eelgrass vary across regions, most recommend the selection of target restoration sites be located in areas where factors associated with eelgrass loss, including disease, are resolved and can be prevented (Evans & Leschen, 2009;Fonseca, Kenworthy, & Thayer, 1998). Ironically, the earliest guidelines for translocation of eelgrass beds following decline due to disease in the Atlantic recommended that transplants be "restricted to bays and estuaries…where waterfowl may be expected to feed" (Cottam & Addy, 1947;p. 397). However, if waterfowl are vectoring eelgrass pathogens during their annual migratory cycles, it is unclear how to prevent the introduction of disease pathogens into restored meadows.

ACK N OWLED G M ENTS
Logistical support was provided by the U. S. Fish and Wildlife Service, Izembek National Wildlife Refuge. Waterfowl cloacal contents were collected by Tyrone Donnelly and Lindsay Carlson under the supervision of Andy Ramey. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

CO N FLI C T O F I NTE R E S T S
Authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
All demultiplexed Illumina MiSeq data can be found at NCBI BioProject PRJNA548352, and sample information will be made publicly available via data release upon publication acceptance. All other data are available in the main text.