Understanding the drivers of population dynamics is a major challenge in population biology, particularly when populations are declining and require mitigation strategies to prevent extinction. Many European farmland birds have declined following farmland intensification and this has been hypothesised as being due to combinations of loss of habitat, shortages of summer and winter food, or a multiplicity of causes [20–22, 43]. Here we add a further potential cause as we show, for the first time, high parasite prevalence in winter, a season associated with significant mortality. This is coupled with evidence suggestive of immunological challenge and a physiological response that may potentially impact upon survival, and we suggest that declining populations may make good systems within which to examine this previously overlooked period in host-parasite ecology. Whilst results here are purely correlative, we discuss the potential implications of our data that warrant further, experimental, work.
The high prevalence of infection found in our population suggests either a high incidence of chronic infections, or stress-induced relapse of existing infections. The vectors of blood parasites tend to be dormant during the winter months, considerably reducing the transmission of new infections . For Haemoproteus parasites, microscopy and PCR techniques tend to provide similar prevalence estimates [44, 45], although Fallon & Ricklefs  found 25-35% of PCR-detected Haemoproteus infections to be undetected through examination of blood smears. Infection intensities within our population are relatively low when compared to other passerine-Haemoproteus spp. systems (e.g. 40–790 parasites per 10,000 RBCs ; means of 6.4 and 12.1 parasites per 10,000 RBCs in the West Indies and Missouri Ozarks respectively ; 90 parasites per 10,000 RBCs ) during the summer months, possibly suggesting that our findings represent the chronic stages of infection, rather than relapses. Unfortunately infection intensity data from our study species during the breeding season is, to our knowledge, only presented in Allander and Sundberg  but this paper measures parasites per 100 microscope fields and does not standardise measurements by erythrocyte abundance, so we are unable to make direct comparisons, although these data from our population would be valuable in assessing the relative effects of parasitism between the breeding and non-breeding periods. However, as the non-infective asexual stages of Haemoproteus reside in the tissues and only infective gametocytes are present in the peripheral circulation [39, 49], this suggests that our findings represent active infections (regardless of whether these are considered to be chronic infections, or relapses) and has further implications for the over-winter ecology of avian blood parasites in temperate species.
Haemoproteus infection was associated with both a reduced H:L ratio, and an increased WBC count. The H:L ratio is generally expected to increase in response to infection , whereas our data suggest the opposite. Similarly to our data however, other systems have also found a reduction in H:L ratio associated with Haemoproteus infection [51, 52], and Galeotti and Sacchi  suggest that the reduction in H:L ratio is explained by the higher concentration of lymphocytes overall, triggered by an increase in immune system activity . This is demonstrated in our data by the increased WBC count in infected individuals, and suggests a direct immune cost of Haemoproteus infection in this species. H:L ratio is also strongly positively correlated with corticosterone levels , but increases with increasing environmental stress, so these data do not support the idea that increased corticosterone, induced by food stress or poor weather, induces relapses of active parasite infection . However, an increase in immune activity also carries costs : for example, in the house sparrow Passer domesticus, Haemoproteus infection reduces immune response to a second immune challenge [54, 55]. Increased immune activity leads to an increased short-term energy consumption which, even at the levels of a 5-15% increase in energy expenditure calculated by Hasselquist & Nilsson  may amplify the impacts of sub-clinical disease at times of food- or weather-related stress, or in populations already under environmental pressure.
We found year-dependent associations between parasite infection and reduced feather length, with no such relationship with skeletal measures of size, suggesting that under certain conditions, parasites can influence feather length through competition for host resources during moult [9, 33], although we acknowledge that this relationship may be mediated by a third, unknown, factor such as body condition during moult. Whilst our parasite data are from the winter months, rather than from months when moult was occurring, the presence of parasites during winter suggests that they were also present during the preceding post-breeding moult (especially if our data represent chronic infections), as post-breeding tends to be the time when infection rates are highest in other systems . In addition, parasite intensities between peak parasite intensity and chronic infections can be correlated  and novel infections between the end of moult and the winter months may be relatively unlikely due to cessation of vector activity, although further work would be required to confirm this. A larger dataset of repeat within-individual data between years would be of great value in elucidating whether this association with wing length occurs at an individual level in relation to infection intensity.
The replacement of year with lowest maximum temperature explained a similar amount of variation in the statistical model, suggesting that temperature may underlie the relationship between parasitism, wing length and year. Parasite prevalence was higher during the second year of the study, coinciding with fewer birds sampled despite similar sampling effort , suggesting fewer birds in the population. This is unlikely to be due to birds moving out of the study area during the second year, as yellowhammers in the UK show very low dispersal  and our population showed a relatively high winter-summer fidelity as estimated through colour-ring re-sightings (Dunn, J. C., Goodman, S. J., Benton, T. G. & Hamer, K. C., unpublished observations). Instead, this may have been due to an early cold spell during autumn 2008 (the second year of the study ), leading to an increase in energy requirements and consequently food requirements, thus resulting in increased over-winter mortality. Given the association between infection and wing length was found only during the first year of sampling when the weather was mild and more birds were caught, and that our survival analysis indicated that birds with longer wings had an enhanced survival probability, it seems plausible that those birds susceptible to physiological effects of parasitism may be more likely to succumb to food stress during a cold winter such as that of the second year of the study and thus be removed from the population. However, this suggestion is speculative and requires further investigation, preferably by inclusion of blood parasite intensity data in the survival analysis.
Farmland birds more generally are under pressure from reduced nesting habitat availability, reduced quality of available nesting habitat, and reductions in both summer and winter food availability [20–22, 43]. Here, we suggest that parasites may also play a significant role in the ecology of declining populations, potentially by interacting with other environmental stressors. These interactions have the potential to occur both within and outside the breeding season, and we suggest that declining populations, especially those such as ours that remain resident outside the breeding season, would make excellent study systems for the examination of such relationships within and between seasons.