To our knowledge this is the first study of the global diversity of contemporary strains of AMDV. The study showed that partial NS1 sequencing can be used for determining major clusters in AMDV outbreaks. The phylogenetic analysis demonstrated substantial genetic diversity among the sequences collected within each country but also within farms, and confirmed results from previous studies of AMDV diversity based on the partial NS1 gene [6, 10,11,12, 18, 19]. A possible explanation for the high genetic diversity found in the present study could be that many of the countries from which the samples originate do not have a defined eradication strategy and have enzootic circulation of AMDV, which increases the risk of introduction of different strains. The relatively high level of nucleotide diversity observed here is in agreement with the generally high mutation rate seen in Parvoviruses [2, 12].
Most viral sequences grouped according to the farm they originated from. However, considerable diversity was also observed within some farms, suggesting multiple introductions of AMDV. This observation is in agreement with a study by Canuti et al.  were 42% of the animals examined for AMDV were infected with more than one viral strain. In countries where no eradication or control measures are carried out, viruses may persist on farms for longer timespans and thereby accumulate higher levels of diversity. The high environmental stability of the virus could compound this problem further. It also seems likely that the fast selection of new breeding stocks from the population within a farm drives the viral evolution even further .
The Danish clusters displayed relatively low levels of diversity probably because all the samples originated from contemporary outbreaks, meaning that several animals were sampled around the same period of time and always in connection with outbreaks. This pattern of low diversity also suggested that a single source was the cause of each of the different Danish outbreaks (Saeby, Holstebro and Zeeland, respectively). The diversity among strains from the three Danish clusters cannot be explained by genetic drift and therefore it is obvious that distinct viral strains caused each of these outbreaks . Samples from the other countries included in this study did not originate from outbreaks but were collected from farms having persistently AMDV infected mink and thus exhibited more diversity. This study furthermore indicated that it might be possible to identify transport of virus between countries. For instance, sequences from Finnish mink clustered close to the North American sequences, in agreement with the idea that Finnish minks were probably initially imported from USA and Canada . Our results also indicated movement of viral strains between Greece and the Netherlands and between Lithuania and Sweden – most likely due to trade of live animals. Furthermore, sequences from American mink clustered close to the Danish Saeby strain, which might be explained by the purchase of live mink from USA in the eighties. The AMDV status of the farms and the mink imported at that time were often not known, and shortly after importing these animals, several farms were found positive for AMDV. Similarly, Canadian sequences were closely related to both the American and Saeby cluster strains, an observation which also can be explained by trade of mink from either USA or Denmark to Canada (Mariann Chriél, personal communication). Unfortunately, there were no farm records available to confirm this.
Of particular interest with respect to the recent Danish outbreaks is that that two sequences from a Swedish farm were closely related to viruses from the outbreak on the Danish island Zealand. The pairwise sequence identity equaled that within the Danish Sæby and Holstebro outbreak clusters, thereby indicating the Zealand outbreak could have an epidemiologic link to Sweden. Zealand had been free of AMDV from 1998 to 2015. Since AMDV is a highly resistant virus it is likely that virus can be transferred between mink farms by trade of live, subclinically infected mink, aerosols between adjacent farms, or through passive transmission with persons, wild life or shared farm equipment [5, 20,21,22].
The sequences originating from the Holstebro outbreak were not closely related to any of the other sequences included in this study and due to their high pairwise sequence identity it was not possible to infer the relationship between them. Thus, the phylogenetic analyses could not provide any additional information on the source of this recent outbreak. The epidemiological data suggested a feed-borne source of the outbreak  and the hypothesis was that one of the feed ingredients was contaminated with AMDV originating from the country where the product was processed. Since a limited number samples were analyzed from each country, it cannot be ruled out that the strains responsible for the outbreak originated from countries included in the study in that most countries have a number of different AMDV strains circulating.
Another possible route through which new strains of AMDV could be introduced in Denmark is via AMDV infected wild mink entering the feed producer’s production facilities or feed silos on farms. Only one wild mink from Denmark and sixteen from Iceland were sampled in this study. The viral sequence obtained from the Danish wild mink was very different from the other known AMDV strains in Denmark, suggesting that the wild population is not the source of the AMDV strains circulating in farmed animals. However, more extensive sampling of the Danish wild mink population would be necessary to more conclusively determine whether the wild mink population acts as a reservoir of AMDV. It should be noted that in addition to mink other members of the mustelid species can also be infected with AMDV  and that they may thereby pose a threat for the farmed mink, and it could therefore be relevant to sample other species as well [18, 19, 23]. Finally, migration of wild mink between Denmark and Germany could also have led to the introduction of new AMDV strains.
Iceland is the only country in the world that has managed completely to eradicate AMDV from mink farms, although one case of AMDV after the eradication suggested that the virus may still be present in the wild population of mink . In the present study, the sequences obtained from the Icelandic wild mink were most closely related to viruses from Sweden, but the sequence identity was not very high. However, the fact that AMDV was present in the wild mink of Iceland, emphasized the importance of having good biosecurity procedures in place to ensure wild mink from gaining access to farm areas, especially in Iceland and other countries where AMDV is circulating in the wild fauna. Conversely, farmed mink may also act as a reservoir for introduction of AMDV into the wild mink population.
It should be emphasized that the section of the NS1 gene analyzed here represents only 7% of the full genome which has the consequence that the present data set provides limited power to detect differences between individual farms during an outbreak, and to track transmission of virus belonging to the same cluster between farms. In addition to the issue of limited resolution of the phylogeny, a drawback with different partial sequences is that they sometimes result in different and conflicting tree topologies [17, 19]. It has however been shown that by sequencing the entire AMDV genome the phylogenetic resolution is increased to a level where transmission routes between farms can be elucidated [24, 25], however the costs associated with the next generation sequencing (NGS) approach are substantial.