Long-distance transmission of pathogenic Vibrio species by migratory waterbirds: a potential threat to the public health

A potential mechanism for the global distribution of waterborne pathogens is through carriage by the migratory waterbirds. However, this mode of transmission has yet been confirmed epidemiologically. Here, we conducted whole genome sequencing of Vibrio spp. collected from waterbirds, sediments, and mollusks in the estuary of the Liaohe River in China to investigate this transmission mode. We found that a V. parahaemolyticus strain isolated from a waterbird was clonally related to the other V. parahaemolyticus strains obtained from the sediments and mollusks, and three V. mimicus strains isolated from bird feces were genomically related to those found in the mollusks and upstream groundwater, suggesting that the bird-carried Vibrio strains were acquired through the direct predation of the local mollusks. Surprisingly, two bird-carried V. parahaemolyticus strains belonging to the same clone were identified in Panjin and Shanghai, which are over 1,150 km apart, and another two were found at two locations 50 km apart, further supporting that waterbirds are capable of carrying and disseminating these pathogens over long distances. Our results provide the first evidence of direct transmission from mollusks to waterbirds and confirm that waterbirds act as disseminating vehicles of waterborne pathogens. Effective surveillance of migratory waterbirds along their routes will be valuable for predicting future epidemics of infectious diseases.


Supplementary text
Appendix 1

Detection of recombination in V. parahaemolyticus (VP)
Previous studies suggested that recombination was the major factor driving the evolution of the V. parahaemolyticus 4 . To remove the potential impact of recombination, RDP program 5 was employed to identify recombination events to obtain a more accurate phylogenetic reconstruction. SNP analysis identified 102,306 V. parahaemolyticus core genome SNPs. As the accessory genomes have been removed from the core genome of VP, the number of recombination events was relatively lower than the values reported by another study 4 .
RDP identified 12 large recombination events in the core genome of VP (Table S5), many of which were also reported previously 4 . Some hotspots of recombination sites, including the genes involved in the ABC transporter substrate-binding protein, secretion protein, DNA binding (peptidases), endonuclease, metalloendopeptidase, and N(6)-L-threonylcarbamoyladenine synthase activities were also identified in this study.
A total of 11,613 recombinant SNPs were removed, which contributed to11.3% of the variants identified in the dataset. For the ST3 strains, a total of 613 recombinant SNPs were removed, which contributed to 0.0468% of the variants identified in the core genome of V. parahaemolyticus. Only non-recombinant core genome SNPs were used to infer the genomic relationship of V. parahaemolyticus. Maximum parsimony (MP) method was used to infer the phylogenetic relationship of ST3 clones by using non-recombinant core genome SNPs. Only one MP tree was generated.

Modeling the dynamics of VP-carrying birds by a predator-prey model
We chose the intertidal zone near the Yingkou City for sampling and modeling, which has a square with approximate 1.5×10 6 m 2 If the prevalence is extremely low (<0.1%), a reasonable approach might be to assume that there is only one positive sample in each positive pool.
Let k be the pool size, x the number of positive pools, and m the number of pools tested. The prevalence of VP-infected mollusk was denoted as ∏.
As suggested by Cowling et al. (1999) 5 , the maximum-likelihood estimator (MLE) of VP prevalence in mollusk (∏) can be written as (1) Note that x (the number of positive pools) has a binomial distribution with parameters: n and P.
The variance of MLE can be estimated by Eq. (2) (2) In our case, the number of pools tested was 138. As all of samples came from six times of samplings (each time consists of 25 sampling points) and average number of mollusk in each sampling point was 660 individuals, the size of the pools was 6×25×660= 99,000. Because six V. parahaemolyticus strains were isolated from mollusks, the number of positive pools was six.
The number of infected of VP-carrying birds can be estimated by the following equations: P denotes the prevalence of VP in the mollusk, while s indicates the daily predation rate. Thus, we can obtain the dynamics of VP infected birds. As we do not know the chance of VP to pass the digestive tract, the predicted number of VP infected birds might be overestimated.