The data presented here show that salmonellosis of wild, particularly passerine birds, in the north of England during 2005 and 2006 was caused mainly by a narrow range of possibly host-adapted S. Typhimurium strains. These strains are generally susceptible to antimicrobials in contrast to many human and livestock-associated strains [12, 13] such as S. Typhimurium DT 104 [12–14]. They contain a range of genes associated with systemic and enteric disease in birds and mammals, but lack sopE, a gene that has been associated with some epidemic strains of S. Typhimurium in both humans and food animals. The sopE gene has been identified in six of 15 outbreak-associated S. Typhimurium phage types, but not in the epidemic multiresistant phage type DT 104, which has been the most common S. Typhimurium phage type in cases of human infection in England and Wales since 1990 .
The majority of Salmonella isolates obtained during this study were of serovar Typhimurium, of which four phage types were identified; DT 40, DT 41, DT 56 and PT U277, with DT 56 being the predominant phage type, widespread spatially and temporally over the north of England (Figure 2). As far as we know, S. Typhimurium DT 56 has not been reported previously in wild birds from outside of the UK, unlike phage types DT 40 and PT U277 that have been reported in wild birds from Scandinavia . It is possible that the predomination of phage type DT 56 may result from the fact that sample collection was geographically and temporally limited to the north of England in 2005 and 2006, therefore a single clone may have been the origin of most of the DT 56 isolates. This hypothesis would be supported by the results of the PFGE analysis carried out by this study where all DT 56 isolates showed 99% similarity to each other. The majority of S. Typhimurium DT 56 and DT 40 isolates were from members of the Fringillidae (greenfinch, goldfinch and siskin) and Passeridae (house sparrow) families, which is consistent with the findings in previous studies [3, 4, 16]. In addition, S. Typhimurium DT56 was isolated from a wood pigeon and a collared dove. S. Typhimurium DT 41 was isolated from starlings and one herring gull: the findings of this study and others would suggest that DT 41 causes only sporadic mortality in 'garden birds' and is more frequently associated with gull and wildfowl species .
S. Typhimurium DTs 40 and 56 were those most often associated with garden bird mortality, as found in previous studies [4, 17, 18]. Pulsed-field gel electrophoresis demonstrated that the S. Typhimurium isolates in this study belonged to a small number of closely related, sometimes clonal, strains. This suggests that S. Typhimurium infection in garden birds is maintained within the wild bird population rather than there being repeated infection of wild birds from other sources. However, we cannot state that transmission never occurs between wild birds and livestock or humans, and wild birds should therefore still be treated as a potential source of Salmonella infection. Indeed, a number of studies carried out in Norway and New Zealand have shown an epidemiological link between wild passerines and human Salmonella outbreaks [11, 19].
The S. Typhimurium isolates clustered into four main PFGE groups which, with three exceptions, were closely correlated to phage-type. An isolate of PT U277 showed 99% similarity to the DT 56 isolates that clustered together in PFGE group 5. PFGE analysis grouped 3 DT 40 isolates into 2 patterns (groups 4 and 6) that were separated geographically (Figure 2). Two DT 41 isolates from live starlings sampled at the same location on the same date shared an identical PFGE pattern with the isolate from a dead herring gull that could not be classified by serotyping, sampled at a different location on a different date. Interestingly, these last three isolates were the only ones to contain the fimbrial associated virulence gene pefA.
PFGE group 5 contained the largest number of isolates, including all of the DT56 isolates obtained during this study, demonstrating that these isolates are members of a clonal strain. Alley et al  reported a similar finding that all DT160 isolates examined by PFGE during an outbreak of salmonellosis in wild passerines and humans in New Zealand were indistinguishable and therefore members of a clonal strain. Due to the small number of isolates phage-typed as DT40 and DT41 in this study, it is not possible to comment on their clonality. S. Typhimurium DT 40 is known to affect wild birds, particularly passerine species , therefore it is important to investigate variation within this phage type further. Similar work carried out on a national scale over a longer time period would be valuable.
Every isolate was found to possess all of the virulence genes that were screened for apart from pefA and sopE. The pefA gene is located on a virulence plasmid rather than the bacterial chromosome . Such plasmids can be serovar-specific, but it has been found that not all isolates of plasmid-bearing serovars contain these plasmids [21, 22]. This may explain the low prevalence of this gene among the Salmonella isolates obtained in the present study. The rest of the virulence-associated genes screened for are located on Salmonella pathogenicity islands (PAIs) 1–5 [21, 23, 24], and are known to be associated with adhesion, cell invasion and intra-cellular survival. These genes, particularly those associated with the Salmonella pathogenicity island 1 (SPI-1) and Salmonella pathogenicity island 2 (SPI-2) type III secretion systems, have been well-characterised for their role in both enteritis and systemic infection in mammalian models. Their roles in avian infection are less clear, though the SPI2 system is a requirement for infection and disease in poultry with the avian specific serovars S. Gallinarum and S. Pullorum [25, 26]. Recently SPI1 and SPI2 have been shown to be involved in both systemic and gastrointestinal tract infection of the chicken by S. Typhimurium. The possession of these virulence factors would suggest that the isolates investigated in the current study have the ability to cause systemic and enteric salmonellosis in their hosts. In addition, the ability of all strains tested to invade and persist in avian macrophage-like HD11 cells, would also suggest that they have the potential to cause systemic infection in avian species: mutant strains of Salmonella incapable of surviving within macrophages in vitro are attenuated for systemic virulence [25, 27].
Interestingly, the Salmonella isolate obtained from a live house sparrow was of the same virulence genotype as the isolates from dead house sparrows and other species. Skyberg et al  found that a number of virulence genes could be detected in both healthy and sick poultry, indicating that some virulence genes and associated PAIs may be widespread in salmonellae isolated from both healthy and sick birds. It may be that other factors, such as nutritional, environmental or physiological stressors are also involved in the development of clinical disease and mortality, or it could be that these healthy birds had only very recently become infected and had not had time to develop disease. Experimental studies have shown that gastrointestinal carriage of Salmonella occurs in passerines after infection; therefore it is possible that healthy birds could be persistent carriers .
None of the isolates examined possessed the sopE gene, which has been found to be present in some strains of S. Typhimurium associated with epidemic disease in both humans and food animals  but not with S. Typhimurium DT 104 . This may indicate that wild passerine strains of S. Typhimurium are generally not involved in the epidemiology of S. Typhimurium infection in humans or livestock in the UK. The absence of antibiotic resistance detected in this study would also support this hypothesis, as many S. Typhimurium isolates from human or production animal sources are resistant to at least one antibiotic [12–14]. It is possible that the S. Typhimurium strains affecting wild passerines are adapted to, and maintained within, the wild bird population. However, this study examined isolates collected only during 2005 and 2006. It would require a more spatially extensive study carried out over a longer time period to gather sufficient evidence to fully support this hypothesis.
The gross post-mortem findings in 20 out of 26 birds examined were similar to those described in previous studies [4, 30–33]. Lesions were most commonly present in the crop of the bird, and it has been suggested that ingluvitis and oesophagitis in passerines with salmonellosis indicates that the crop and oesophagus are predilection sites for bacterial invasion. It is possible that progression of this infection to systemic salmonellosis involving other organs (liver and spleen) may occur if the bird's immune system became further compromised . No gross lesions were noted in two dead birds from which Salmonella was isolated, a greenfinch and a Eurasian siskin. In these cases where Salmonella appeared as an incidental finding, it is impossible to know if these birds would have succumbed to salmonellosis in the future (had they not been killed by some other cause) or for how long the bird had been infected before death. Studies have shown that chickens can carry Salmonella enterica containing virulence genes asymptomatically , and this must therefore be a possibility in wild birds also. The only way to determine if this is the case would be through a longitudinal study of individual wild birds.