|Home | About | Journals | Submit | Contact Us | Français|
Low pathogenic avian influenza (LPAI) viruses are well documented to circulate within wild waterfowl populations (Olsen et al. 2006). It has been assumed that these infections are benign with no subsequent effects on life-history parameters. The study by Latorre-Margalef et al. (2009; hereafter L.-M. et al.) represents an important step, as they attempt to test this assumption in wild birds. L.-M. et al. captured migrating mallards (Anas platyrhynchos) at a staging area and tested them for the presence of avian influenza A virus (IAV). They related IAV infection status to body mass and duration of time spent on the staging area. Overall, the study is well designed with impressive sample sizes and the analyses are carefully conducted and presented. However, in discussing these results, the authors assume causation based upon correlation and, although they acknowledge the possibility of immunosuppression during migration due to reduced energy stores, they do not discuss it as a possible explanation for their findings. Below, we consider several of the major findings by L.-M. et al., providing alternative explanations for the results. Because the L.-M. et al. study design is correlational, it is not possible to use their data to distinguish between their interpretations and our alternative explanations.
L.-M. et al. used data from the first capture of birds in a particular year to examine variation in body mass in relation to IAV infection. They ignored data from recaptured individuals and controlled for other factors known to influence body mass (i.e. sex, age, structural size) before testing for variation related to IAV infection. In Results, the authors state ‘Body mass was lower in the positive ducks compared with the negative ducks when other variation in body mass was controlled for’. This statement accurately reflects the statistical result. However, in the Discussion, they summarize this result as: ‘Mallards with LPAI infection did indeed lose more weight than healthy birds’. Furthermore, they state ‘…LPAI infection affected the body mass of Ottenby mallards…’.
The conclusion by L.-M. et al. assumes that all birds started at an equal mass and those contracting IAV lost more mass, resulting in the difference at the time of capture. Infection with IAV has been shown to affect body mass (Gharaibeh 2008) and, although emaciation has been reported in association with mildly pathogenic IAV, it is infrequent because of the acute nature of IAV infections (Swayne & Halvorson 2003). Thus, it is possible that IAV infection causes a reduction in body mass of wild birds. However, the analyses provided by L.-M. et al. are based upon a single data point for each bird and therefore cannot be used to infer that mass loss results from IAV infection.
An alternative explanation is that birds in poorer condition are more susceptible to IAV infection. This alternative assumes that birds with a lower mass (for any reason) are more likely to be IAV positive at capture and is supported by the relationship between condition, a measure of which is body mass, and immunocompetence. Hanssen et al. (2005) found that incubating female common eiders (Somateria mollissima) that lost more body mass had reduced immune function, as measured by lymphocyte levels and antibody response to two antigens. Bourgeon et al. (2006) found that the phytohaemagglutinin (PHA) response and immunoglobulin index decreased along with body mass during incubation in common eiders. Navarro et al. (2003) reported that house sparrows (Passer domesticus) in good body condition had stronger PHA responses than those in poor condition. Owen & Moore (2008) studied four species of thrushes, finding that birds in poor energetic condition had low leucocyte and lymphocyte counts. Accordingly, the association of poor body condition with decreased immune response leads to the alternative hypothesis that body mass affects the likelihood of IAV infection.
L.-M. et al. also used the RT-PCR Ct-value, described as ‘…inversely proportional to the number of initial copies of the RNA template’, as an index to the severity of IAV infections. The logic was that a low Ct-value is an indication of greater virus shedding. The results are only significant for juvenile birds and are stated as: ‘the birds shedding less viruses (high Ct-values) had higher body mass than those shedding more viruses (low Ct-values)…’ Again, this statement accurately reflects the results of these analyses. In the Discussion, these results are inferred to demonstrate that their previous conclusion (i.e. LPAI infection leads to greater mass loss) ‘…seems very robust since infection status…as well as the Ct-value…was related to body mass’. Alternatively, IAV-infected birds with low body mass may shed more viruses than infected birds with higher body mass because reduced immune function inhibits the ability to combat the infection. As such, body condition may influence the severity of IAV infection.
L.-M. et al. used data from recaptured individuals to test for differences in residence time at the Ottenby staging area in relation to IAV infection status at first capture. Again, the results applied only to juveniles and are clearly stated as: ‘…there was a negative effect of the Ct-value in September only, i.e. the less virus the mallards shed, the shorter their staging period’. In the Discussion, they refer to this result as support for their conclusions regarding the effect of IAV infection on body mass loss.
This conclusion fails to consider that duration of time spent on staging areas may be directly related to body condition. Mallards typically migrate in intervals of two to three days during which they metabolize stored reserves with periodic stops at staging areas where they stay one to four weeks as reserves are replenished (Yamaguchi et al. 2008). Thus, birds in poorer body condition probably stage longer to rebuild reserves for the next step of migration. Under such a scenario, the relationship between virus shedding and staging time may be spurious, caused by correlation of both staging time and Ct-value with body mass. As such, the longer staging time for birds shedding high amounts of viruses may be a direct effect of body condition per se, not IAV infection.
L.-M. et al. make several statements meant to imply cause-and-effect relationships between IAV infection and adverse effects on wild mallard ducks. The most obvious of these is the title of the paper. However, because their study is quasi-experimental in design (i.e. where study subjects are not randomly assigned to treatment categories), L.-M. et al. have a very limited ability to infer cause-and-effect relationships (Cook & Campbell 1979). We have no data to demonstrate that our alternative explanations are correct while those of L.-M. et al. are incorrect; we only argue that they are equally plausible. We agree that correlations exist between body condition and the probability, as well as severity, of IAV infection. However, we argue that, based on findings reported by L.-M. et al., it is not possible to determine whether IAV infection influences body condition or vice versa. Thus, if our alternative explanations are plausible, then the reported effects of IAV infection on migrating mallard ducks by L.-M. et al. are equivocal. We suggest that further work is necessary to adequately test the effects of low pathogenic IAV infection on life-history parameters of hosts.
The accompanying reply can be viewed at http://dx.doi.org/doi:10.1098/rspb.2009.0275.