INTRODUCTION
Most avian species inhabiting temperate regions breed in spring and summer. In these species, the prevalence of blood parasites is generally higher during than outside the reproductive period (Weatherhead and Bennett, Reference Weatherhead and Bennett1991; Rintamaki et al. Reference Rintamaki, Huhta, Jokimaki and Squires-Parsons1999; Deviche et al. Reference Deviche, Greiner and Manteca2001; Garvin and Greiner, Reference Garvin and Greiner2003; Murata et al. Reference Murata, Tamada, Ichikawa, Hagihara, Sato, Nakamura, Nakamura, Sakanakura and Asakawa2007; Cosgrove et al. Reference Cosgrove, Wood, Day and Sheldon2008; Hartup et al. Reference Hartup, Oberc, Stott-Messick, Davis and Swarthout2008), but the factors that account for this difference and their relative contribution to haematozoan infections remain debated.
One such factor is the abundance of appropriate haematozoan vectors (Klei and DeGiusti, Reference Klei and DeGiusti1975; Bennett et al. Reference Bennett, Thommes, Blancou and Artois1982; Godfrey et al. Reference Godfrey, Pence and Fedynich1990; Earle et al. Reference Earle, Horak, Huchzermeyer, Bennett, Braack and Penzhorn1991). For example, geographical differences in infection with Haemoproteus chloris between populations of Greenfinches, Carduelis chloris, parallel putative differences in the abundance of the vectors for this parasite (Merilä et al. Reference Merilä, Bjorkund and Bennett1995). Similarly, differences in haematozoan (Leucocytozoon spp., Haemoproteus spp., Plasmodium spp. and Trypanosoma spp.) infection between mainland and island populations of passerines are associated with the presence or absence, respectively, of appropriate vectors (Super and van Riper, Reference Super and van Riper1995). If vector abundance plays a primary role in determining haematozoan prevalence through new infections, a seasonal increase in vector abundance is predicted to be temporally associated with an increase in patent haematozoan infections.
A second factor is the breeding effort. This effort is associated with an elevated blood heterophil/lymphocyte (H/L) ratio (Ots and Hõrak, Reference Ots and Hõrak1996), which increases in response to chronic stress or administration of the stress hormone corticosterone (Gross and Siegel, Reference Gross and Siegel1983; Gross, Reference Gross1988; Ilmonen et al. Reference Ilmonen, Hasselquist, Langefors and Wiehn2003; Ots and Hõrak, Reference Ots and Hõrak1996). These observations suggest that the breeding effort constitutes a major stressful event in the annual cycle with the potential to decrease resistance to new parasitic infections and/or precipitate a relapse of latent infections. Consistent with this contention, avian studies found a trade-off between reproductive investment and resistance to haemoparasitic infections (Leucocytozoon sp.: Norris et al. Reference Norris, Anwar and Read1994; Haemoproteus sp.: Ots and Hõrak, Reference Ots and Hõrak1996; Haemoproteus majoris: Allander, Reference Allander1997; Gustafsson et al. Reference Gustafsson, Nordling, Andersson, Sheldon and Qvarnström1994), and avian as well as mammalian studies indicate that the activity of the reproductive system can influence these infections (Bensch et al. Reference Bensch, Waldenstrom, Jonzen, Westerdahl, Hansson, Sejberg and Hasselquist2007; Wood et al. Reference Wood, Cosgrove, Wilkin, Knowles, Day and Sheldon2007; Arriero et al. Reference Arriero, Moreno, Merino and Martinez2008; Fokidis et al. Reference Fokidis, Greiner and Deviche2008). If the seasonal breeding effort is, indeed, physiologically stressful, it is predicted in previously infected birds that it will be associated with increased haematozoan prevalence resulting from relapse and/or parasite reproduction, independent of vector abundance.
The gonadal hormone testosterone (T) is also thought to influence haematozoan infections. Plasma T in many species is seasonally higher in males than females (Deviche and Gulledge, Reference Deviche and Gulledge2000; Ketterson et al. Reference Ketterson, Nolan and Sandell2005; Spinney et al. Reference Spinney, Bentley and Hau2006) and this difference parallels a sex difference in blood parasite prevalence (Kiyota et al. Reference Kiyota, Korenaga, Nawa and Kotani1984; Zuk and McKean, Reference Zuk and McKean1996, but see Hartup et al. Reference Hartup, Oberc, Stott-Messick, Davis and Swarthout2008 and Pawelczyk et al. Reference Pawelczyk, Gryczynska and Mazgajski2003). In addition, adult male Dark-eyed Juncos, Junco hyemalis, that were undergoing a first reproductive cycle had lower plasma T and also lower haematozoan prevalence than older males (Deviche et al. Reference Deviche, Greiner and Manteca2001), and T administration to free-ranging males of this species increased their blood parasite prevalence (Deviche et al. Reference Deviche, Parris and Greiner2006). In the Large Vesper Mouse, Callomys callosus, inoculation with Trypanosoma cruzi resulted in lower blood prevalence of this parasite in castrated than control animals and T treatment to castrates increased their parasite prevalence to control levels (Do Prado et al. Reference Do Prado, Levy, Leal, Bernard and Kloetzel1999). Given to House Mice, Mus musculus, T treatment increased the survival of the intestinal nematode, Heterakis spumosa (Harder et al. Reference Harder, Wunderlich and Marinovski1992) and it enhanced susceptibility to new parasitic infections with the intestinal roundworm, Strongyloides ratti (Kiyota et al. Reference Kiyota, Korenaga, Nawa and Kotani1984). If elevated plasma T stimulates re-emergence of tissue-bound parasites in the blood, for example secondary to immunosuppression, periods of elevated plasma T in males are predicted to be associated with an increase in patent infections, even in absence of appropriate vectors.
In previous studies using free-ranging birds, the relative contribution of the above factors to blood parasite prevalence was difficult to assess because most species reproduce and males have seasonally elevated plasma T at the same time of the year that vectors are most abundant (Garvin and Greiner, Reference Garvin and Greiner2003; Rintamaki et al. Reference Rintamaki, Huhta, Jokimaki and Squires-Parsons1999). To address this issue, we investigated seasonal changes in blood parasite prevalence and diversity in a population of White-winged Crossbills, Loxia leucoptera. Crossbills have a specialized diet consisting primarily of conifer seeds, which are available irregularly both geographically and temporally (Bent, Reference Bent1968; Newton, Reference Newton1972; Benkman, Reference Benkman, Poole and Gill1992). To accommodate this irregularity, crossbills have evolved flexible reproductive cycles that can extend from January to November (Benkman, Reference Benkman, Poole and Gill1992; Deviche and Sharp, Reference Deviche and Sharp2001). Crossbills in the present study began to breed in winter and plasma T in males was seasonally elevated in late winter and early spring, when ambient temperatures were well below freezing and haematozoan vectors absent (Deviche and Sharp, Reference Deviche and Sharp2001; Werner, Reference Werner1992). By summer, when putative vectors became abundant, birds had completed their breeding cycle and T in males was circulating at baseline concentrations (Deviche and Sharp, Reference Deviche and Sharp2001; see Results section). We reasoned that if the winter breeding effort or a seasonal increase in plasma T (in males) is associated with increased blood parasite prevalence, as predicted by other studies, this increase must result from an increase in reproduction (Trypanosoma sp.) or a relapse (Haemoproteus sp. and Leucocytozoon sp.; see Results section) of parasites rather than new inoculations. By contrast, if a seasonal increase in haematozoan vector abundance is primarily responsible for controlling seasonal changes in blood parasitaemia through new infections, prevalence was predicted not to increase until the spring emergence of vectors.
To further investigate the control of haematozoan parasite infections, we compared seasonal changes in parasite prevalence in adult males and females and determined whether juvenile birds can be infected with these parasites. To determine whether the late winter breeding period was associated with chronic stress, we determined the body condition of birds sampled during this season, measured the blood H/L ratios of these birds, and compared the results with those from birds sampled in summer, after completion of breeding. We hypothesized that if birds in winter-early spring are exposed to chronic stress resulting from environmental factors or breeding effort, these birds would be in lesser body condition and have higher H/L ratios than summer birds.
Few studies have analysed changes in haematozoan prevalence throughout the year in free-ranging passerines (Blue Tit, Cyanistes caeruleus: Cosgrove et al. Reference Cosgrove, Wood, Day and Sheldon2008; Rufous-winged Sparrow, Aimophila carpalis: Deviche et al. Reference Deviche, McGraw and Greiner2005; House Finch, Carpodacus mexicanus: Hartup et al. Reference Hartup, Oberc, Stott-Messick, Davis and Swarthout2008) and other birds (Rock Pigeon, Columba livia: Klei and DeGiusti, Reference Klei and DeGiusti1975; Japanese Rock Ptarmigan, Lagopus mutus japonicus: Murata et al. Reference Murata, Tamada, Ichikawa, Hagihara, Sato, Nakamura, Nakamura, Sakanakura and Asakawa2007; Helmeted Guineafowl, Numida meleagris: Earle et al. Reference Earle, Horak, Huchzermeyer, Bennett, Braack and Penzhorn1991; Red-bellied Woodpecker, Melanerpes carolinus: Schrader et al. Reference Schrader, Walters, James and Greiner2003). A better understanding of the factors that control avian haematozoan infections requires additional seasonal studies using species that inhabit a diversity of habitats and are, therefore, exposed to a variety of environmental conditions during and outside their breeding season. To our knowledge, the present work is the first to describe changes in blood parasitaemia throughout the year in a subarctic passerine.
MATERIALS AND METHODS
Bird capture and sampling
We caught 131 adult male and 102 adult female White-winged Crossbills at the same locality in Fairbanks (Alaska, USA; 65°52′N/147°54′W) between 3 May 1998 and 6 August 1999, using Japanese mist nets and a live decoy bird placed next to the net. This sample consisted of 226 first captures (126 males, 100 females) and 7 recaptures (5 males, 2 females). Data from recaptured birds were considered independent from first capture data and were retained in the analyses as long as birds were recaptured at least 1 month after the first capture. Starting in 1999, adults (=after-hatch-year, AHY) were separated into 2 age classes (second-year, SY: n=104; after-second-year, ASY: n=69; unknown: n=4) based on plumage characteristics (Deviche, Reference Deviche2000; Pyle, Reference Pyle1997). Birds were sexed based on plumage colour and previously described morphological characteristics (developed cloacal protuberance in males; incubation patch in females: Pyle, Reference Pyle1997).
To determine whether hatch-year (HY) birds can be infected during their first summer, we analysed blood parasites in samples collected from birds of this age class caught in July and August 1998 (n=11) and 1999 (n=19). When sampled, HY crossbills were no more than a few months old as determined by incomplete skull pneumatization (Pyle, Reference Pyle1997).
Birds were caught between 6:20 a.m. and 4:10 p.m. (Alaska Standard Time). All activities were pre-approved by the University of Alaska Fairbanks Institutional Animal Care and Use Committee and were conducted with appropriate scientific collecting permits from the Alaska Department of Fish and Game, US Fish and Wildlife Service, and the US Geological Survey Bird Banding Laboratory.
Within 5–10 min of capture, we removed birds from the mist net and collected a blood sample (max. 700 μl) from a wing vein as described by Deviche and Sharp (Reference Deviche and Sharp2001). Five to 10 μl of blood were used to immediately prepare 1 or 2 thin blood smears on 25×75 mm glass microscope slides (Bennett, Reference Bennett1970). Smears covered, on average, a surface equal to 500–800 mm2. The remainder of the blood was used to measure circulating concentrations of reproductive hormones, including T, in males (Deviche and Sharp, Reference Deviche and Sharp2001). Slides were dried at room temperature, dehydrated, stained using the Giemsa method, and cover-slipped as described by Deviche et al. (Reference Deviche, Parris and Greiner2006). Representative slides with smears containing the 3 parasite species under study (see Results section) were deposited at the US National Parasite Collection (Beltsville, MD; Accession numbers: USNPC 102020-102023). After blood collection, we measured the wing chord of each bird to the nearest mm, weighed birds to the nearest 0·1 g, corrected these weights for the volume of blood taken, marked birds with an individually numbered aluminum tarsal band (US Geological Survey, Washington, DC), and released them at the capture site.
Blood smear analysis
Representative smears were studied using light microscopy at 1000×magnification under oil immersion to identify parasite species. Parasites were identified to species level using morphological criteria established by Peirce and Baker (Reference Peirce, Baker, Lumsden and Evans1976) and Valkiunas (Reference Valkiunas2005). For quantification, each smear was examined first at low (200×; 10 min) and then at high (400×; 5 min) magnification and birds were classified as negative or positive for each haematozoan species (see Results section; Deviche et al. Reference Deviche, Greiner and Manteca2001). We defined observed prevalence (hereafter prevalence) as the proportion (%) of sampled birds in which we detected blood infection with at least 1 individual of a particular parasite species (Bush et al. Reference Bush, Lafferty, Lotz and Shostak1997). We also measured the abundance of Leucocytozoon fringillinarum (hereafter L. fringillinarum), defined as the number of these parasites per 100 visual fields of a smear studied at 400× magnification (Bush et al. Reference Bush, Lafferty, Lotz and Shostak1997; Deviche et al. Reference Deviche, Greiner and Manteca2001). Given the area of individual circular visual fields at this magnification (diameter: 50 μm; area: 2×10−3 mm2) and the smear area (500–800 mm2; see above), abundance data could be obtained from non-overlapping fields. Parasites other than L. fringillinarum were generally present in low numbers and their abundance was not measured (Levine, Reference Levine1985). Smears were analysed without knowledge of sex, age, or sampling date.
To determine the reliability and consistency of parasite detection, 25 randomly chosen smears, each from a different bird, were studied as described above by 2 observers (P.D.; H.B.F.) who independently measured the abundance of L. fringillinarum and the prevalence of Trypanosoma avium and Haemoproteus fringillae (hereafter T. avium and H. fringillae). In addition, each observer measured these parameters in pairs of smears prepared from a same blood sample of 25 randomly chosen birds.
We compared prevalence in summer-caught HY crossbills (n=30; see above) and AHY birds caught during the same period (1998: n=11 males and 7 females; 1999: n=8 males and 1 female). The proportion of HY birds during the considered period was higher in 1999 (HY/AHY: 2.1) than 1998 (id., 0·6; Fisher exact probability test: P=0·034). In addition, HY crossbills in July and August had not begun to moult into first pre-basic plumage and their sex could not be determined. The sex ratio of the sampled adult population (n=231 birds, see above) was 43% females. For blood parasite data in HY and AHY birds to be as comparable as possible, it was necessary that (a) the proportion of HY to AHY birds be similar in 1998 and 1999 and (b) the assumed sex ratio of the HY sample be similar to that of the AHY sample (43%). This was achieved by excluding smears from 10 randomly chosen adult males and 2 randomly chosen adult females caught in 1998 from the data set before statistical analysis. Excluding these birds resulted in an AHY sample consisting of 1 male and 5 females (1998) and 8 males and 1 female (1999), for a combined total of 15 AHY birds. Using this approach, HY/AHY ratios were similar in 1998 (11/6) and 1999 (19/9), and so were the assumed HY and effective adult sex ratios (6/15=40% females).
Blood leucocyte profiles
We compared the blood abundance of 2 types of leucocytes (heterophils and lymphocytes) in adults sampled in summer 1998 (10 males, 10 females; average sampling date: 30 July), when birds had the highest seasonal average prevalence of blood parasites (see Fig. 1) with that in winter 1999 (11 males, 15 females; average sampling date: 19 March), when males had seasonally elevated plasma testosterone and birds of both sexes had low prevalence of blood parasites (Fig. 1). For this, we counted the number of heterophils (H) and lymphocytes (L) seen in non-overlapping microscope visual fields of smears studied at 400×magnification and for a constant duration (40 min per smear). Abundance data were used to calculate H/L ratios. Ratios were used in statistical analyses only when the corresponding total number of immune cells (H+L) was at least 30 (n=32 birds).
Fig. 1. (A) Seasonal changes in average daily temperatures (means±s.e.m.) and male crossbill plasma testosterone (T; medians+0·5 interquartile intervals) at the same sampling location in Fairbanks, Alaska. (B–C) Seasonal changes in prevalence of (b) Leucocytozoon fringillinarum, (c) Haemoproteus fringillae, and (d) Trypanosoma avium measured in thin blood smears from adult male and female White-winged Crossbills, Loxia leucoptera. Numbers at the bottom of panel B columns indicate sample sizes. Sample sizes are the same for the other panels. * Indicates an intersexual difference (P<0·05 and non-overlapping 95% confidence intervals) in prevalence.
Body condition
We used the same adults as used for leucocyte profile analysis to determine if body condition is sex dependent and differs between winter and summer. Body condition, a measure of stored energy reserves, was calculated using the residuals of a linear regression between wing chord and body mass (Costantini et al. Reference Costantini, Casagrande and Dell'omo2007; Ochs and Dawson, Reference Ochs and Dawson2008). It was assumed that birds with a larger body mass relative to their body size (estimated by wing chord) were in better body condition.
Ambient temperatures
Daily minimum and maximum temperatures during the study period were obtained from the Alaska Science Center (http://alaska.usgs.gov/science/biology/index.php) and used to calculate average daily temperatures.
Testosterone assay
Plasma T concentrations in samples collected from adult males were measured using a commercial, validated radioimmunoassay (Deviche and Sharp, Reference Deviche and Sharp2001). Hormone concentrations were not measured in samples obtained from females or HY birds.
Statistical analyses
We used χ2 tests to measure associations between parasite species in smears obtained from adult birds. Analysis of these associations required 6 independent χ2 tests (see Results section). A Bonferroni correction was accordingly applied and differences were considered statistically significant when corresponding probabilities were equal to or less than 0·008 (=0·05/6).
We used Pearson product moment correlations, McNemar tests for the significance of change, and Wilcoxon matched-pair signed-rank tests (Siegel, Reference Siegel1956) to determine the consistency and repeatability of parasite detection between observers and between pairs of smears collected from the same bird. Statistical associations between various independent variables (sex, age, plasma T, and average daily temperature on capture day) and prevalence of each parasite species were analysed by multiple logistic regressions.
Binary comparisons of prevalence between sexes at different times of the year (see below) and between age groups (SY vs ASY; HY vs AHY) were done using Fisher exact probability tests. When a difference in prevalence was statistically significant (P⩽0·05), we used a confidence interval for proportion calculator (http://www.dimensionresearch.com/resources/calculators/conf_prop.html) to calculate the 95% confidence interval for each group of data considered. Differences were considered biologically meaningful and are discussed as such only if a Fisher exact probability test was statistically significant and confidence intervals calculated for the two corresponding groups of data did not overlap.
To examine seasonal changes and sex differences in adult blood parasite prevalence and diversity, data (131 males and 100 females; see above) were sorted by capture date and then divided into consecutive, non-overlapping blocks of data, each consisting of at least 10 males and 10 females. Prevalence in males and females within each data block was compared statistically as described above.
Seasonal (winter vs summer) and sex differences in immune function (H and L abundance; H/L ratios), wing chord, body mass, and body condition were analysed by two-way analysis of variance with season and sex as independent variables. Body mass data were not normally distributed and were, therefore, ranked before analysis (Conover and Iman, Reference Conover and Iman1981).
RESULTS
Parasite identification and overall prevalence
Blood smears were found to contain 3 protozoan parasite species: L. fringillinarum, T. avium, and H. fringillae. Forty-eight percent of adult birds sampled were found to contain at least 1 parasite species. The most prevalent parasite species was L. fringillinarum (42% of the samples). T. avium and H. fringillae had prevalences of 13%. The percentages of adult birds sampled in which we observed 1, 2, or 3 parasite species were 34%, 10%, and 5%, respectively.
Inter- and intra-observer consistencies
Data from 2 observers independently measuring the abundance of L. fringillinarum in the same smears were correlated (r2=0·995, P<0·0001) and did not differ (Wilcoxon test: P=0·206). These observers detected T. avium and H. fringillae with similar effectiveness (McNemar tests: Ps >0·05). L. fringillinarum abundance measured in paired samples obtained from the same birds was consistent (r2=0·962, P<0·0001) and there was no within-pair difference in T. avium or H. fringillae detectability (McNemar tests: P>0·05). Thus, the method used to detect blood parasites in smears and to quantify prevalence was reliable and consistent.
Associations between parasite species
In adults, no association was detected between infections with H. fringillae and L. fringillinarum. Indeed, the proportions of blood samples containing L. fringillinarum did not differ whether these samples contained or not H. fringillae (P=0·037; critical statistical significance level=0·008: see Materials and Methods section and Table 1). Likewise, the likelihood of blood infection with H. fringillae did not differ whether birds were or were not infected with L. fringillinarum (P=0·037).
Table 1. Observed frequency of blood infection (% birds infected) with one parasite species in the absence (Neg.) or presence (Pres.) of another parasite species in smears obtained from 231 adult White-winged Crossbills, Loxia leucoptera
* P<0.001 (χ2 test) for comparison of Neg. with corresponding Pos. data.
n.s., Not significant (P>0.008; χ2 test).
By contrast, blood infection with H. fringillae was associated with increased probability of infection with T. avium. Furthermore, infection with T. avium was associated with increased probability of infection with H. fringillae or L. fringillinarum, and infection with L. fringillinarum was associated with increased probability of infection with T. avium (Table 1).
Seasonal changes in ambient temperature and adult haematozoan prevalence
Average daily temperatures varied between −24·4±8·2°C (January 1999) and 17·8±2·2°C (June 1999; Fig. 1A). These temperatures stayed below freezing from October 1998 to 1 April 1999 (except 25 March 1999: +1°C). The minimum daily temperature in 1999 did not reach 0°C until 19 April. Maximum daily temperatures never exceeded 0°C between 26 October 1998 and 14 March 1999.
The prevalence of L. fringillinarum (male and female data combined, n=231) changed seasonally and was positively associated with average daily temperatures (multiple logistic regression: Ps<0·001; Fig. 1A, B). This association was observed also when male (n=131) and female (n=100) data were analysed separately (P's<0·001). Only January samples were found not to contain L. fringillinarum. By March the prevalence of this parasite in males had increased to 18% and by early April it was 27% in males and 42% in females. The late winter and early spring increase in prevalence took place as ambient temperatures were below freezing (Fig. 1A). The highest prevalence of L. fringillinarum occurred during the summer, when ambient temperatures were highest and putative vectors most abundant. Inclusion of sex as factor in the logistic regression models revealed that males had, on average, higher L. fringillinarum prevalence than females (P=0·029). Pairwise comparisons (Fisher exact probability tests followed with confidence interval calculation) revealed higher prevalence of this parasite in males than females only in May and early June 1999 (Fig. 1B).
The prevalence of T. avium and H. fringillae was also positively associated to average daily temperatures (multiple logistic regressions, n=231, Ps<0·001; Fig. 1C, D). These parasites either were not found in blood (H. fringillae) or were found at low (⩽10%) frequency (T. avium) until May (Fig. 1C, D). In contrast to the situation for L. fringillinarum, the prevalence of T. avium and H. fringillae was not sexually dimorphic (Ps>0·60).
Plasma testosterone
Plasma T in adult males changed seasonally (Kruskal-Wallis analysis of variance on ranks: H=23·6, P=0·003; Fig. 1A). Plasma T was low in October 1998, increased between that month and March 1999, when peak levels were measured and average ambient temperatures were well below freezing. Plasma T then decreased and remained low (average: 0 ng/ml) for the rest of the sampling period. Slightly elevated levels in July 1998 did not differ significantly from levels measured in April–June 1999.
Association between adult male plasma T and haematozoan prevalence
We used data obtained from adult males in which plasma T concentrations were measured (1998: n=33; 1999: n=85) to determine whether plasma T was related to parasite prevalence. Multiple logistic regression analyses with ambient temperature and plasma T as independent variables revealed no statistical relation between plasma T and prevalence of any parasite species (Ps>0·60; Fig. 1).
Age differences in adult haematozoan prevalence
Data collected in 1999 were used to compare prevalence in SY and ASY birds. For each parasite species, we performed a multiple logistic regression with age, sex, and ambient temperature as independent variables. Besides confirming the above-described positive associations between prevalence of each parasite species and temperature, these analyses revealed that overall, SY birds had lower H. fringillae blood prevalence than ASY birds (P=0·002; Fig. 2). Separate analyses of data obtained for male and female belonging to each age class revealed a lower prevalence of this parasite in SY than ASY females (P=0·006; Fig. 2). The prevalence of H. fringillae in SY and ASY males also differed statistically (P=0·042; Fig. 2), but this difference was not considered biologically meaningful because confidence intervals overlapped. Multiple logistic regressions found no age-associated difference in adult prevalence of L. fringillinarum or T. avium.
Fig. 2. Prevalence of (a) Leucocytozoon fringillinarum, (b) Haemoproteus fringillae, and (c) Trypanosoma avium in blood smears of second year (SY) and after second year (ASY) male and female White-winged Crossbills, Loxia leucoptera, sampled in 1999 at a single location in Fairbanks, Alaska. Numbers at the bottom of the upper panel columns indicate sample sizes. * Indicates a significant age difference in prevalence.
Adult blood immune cells and haematozoan prevalence
The abundance of blood heterophils and lymphocytes was higher in summer than in winter (heterophils: F1,42=37·1, P<0·001; lymphocytes: F1,42=22·9, P<0·001), but it was not sexually dimorphic and there was no season×sex interaction (Ps >0·1; Fig. 3). The abundance of heterophils was positively related to that of lymphocytes in winter (Pearson product moment correlation coefficients: r=0·495, P=0·01) as well as summer (id.: r=0·835, P<0·001). As a result of this relation, H/L ratios did not differ between seasons or sexes and there was no season×sex interaction for this parameter (Ps>0·2).
Fig. 3. Abundance (number of cells counted on non-overlapping microscope fields of blood smears studied at 400×magnification for 40 min; means±S.E.) of heterophils and lymphocytes during winter and summer in adult male and female White-winged Crossbills, Loxia leucoptera sampled at a single location in Fairbanks, Alaska. Numbers at the bottom of the upper panel columns indicate sample sizes. * Denotes a seasonal difference (P<0·05, two-way analysis of variance).
The low prevalence of blood parasites during winter (Fig. 1) did not warrant an analysis of relations between blood parasites and immune cell abundance during that season. In summer, the abundance of L. fringillinarum was positively correlated to that of heterophils (n=20; Pearson product moment correlation coefficient=0·53, P=0·016) but not lymphocytes (id., P>0·4). By contrast, there was no relation between the prevalence of T. avium or H. fringillae and the abundance of immune cells. This abundance did not differ whether smears were positive or negative for either parasite species (Student's t-tests: P>0·4).
Adult body size and condition
Males had a longer wing chord than females (males (n=24): 87±0·3 (S.E.) mm; females (n=25): 82±0·4 mm; F1,45=74·5, P<0·001), but wing chord did not differ between winter and summer and there was no sex×season interaction for this variable (Ps>0·5).
Males were on average heavier than females (F1,45=12, P=0·001) and birds were heavier in winter than summer (F1,45=6·7, P=0·013; summer males: 27±0·5 (S.E.) g; winter males: 28±0·3 g; summer females: 25±0·3 g; winter females: 27±0·7 g). Crossbills were in better body condition in winter than summer (two-way analysis of variance of residuals of linear regression of wing chord over body mass: F1,45=7·18, P=0·01), but body condition did not differ between sexes and there was no season×sex interaction for this variable (Ps>0·5).
Haematozoan prevalence in hatch-year birds
Prevalence of L. fringillinarum and H. fringillae was lower in hatch-year birds than in adults (=AHY; Fisher exact probability tests: Ps<0·001; no confidence interval overlap; Fig. 4). Prevalence of T. avium did not differ between the two age groups (Fisher exact probability test: P=0·09).
Fig. 4. Prevalence of Leucocytozoon fringillinarum, Haemoproteus fringillae, and Trypanosoma avium in blood smears of hatch-year (HY) and adult (AHY) White-winged Crossbills, Loxia leucoptera, sampled in 1998 and 1999 at a single location in Fairbanks, Alaska. * Denotes an age difference.
DISCUSSION
The goals of the present study were to identify haematozoa in White-winged Crossbills, investigate seasonal and sex-and age-related changes in prevalence of these parasites, and identify factors that potentially account for these changes. Crossbills in the study population bred and males had high plasma T during winter and early spring (Deviche and Sharp, Reference Deviche and Sharp2001). We hypothesized that if haematozoan prevalence increased at these times, this increase presumably resulted from re-emergence of parasites (i.e. increase in reproduction (T. avium) or relapse (H. fringillae and L. fringillinarum)) in the blood and not from new inoculations. Based on previous studies, we also hypothesized that this re-emergence may be related to the breeding effort and/or (in males) elevated plasma T. In late spring and summer, as ambient temperatures increased and putative vectors emerged, birds completed their breeding cycle and plasma T secretion in males had decreased to seasonally low to minimum levels (Deviche and Sharp, Reference Deviche and Sharp2001; see Results section). At this time, an increase in haematozoan prevalence would presumably result from increased exposure to vectors causing new infections, as other avian studies suggest (see Introduction section). The findings are consistent with the conclusion that, depending on the time of year, seasonal changes in blood parasite prevalence in crossbills involve re-emergence of these parasites in the blood or new infections.
White-winged Crossbill blood samples contained 3 parasite species: L. fringillinarum, T. avium, and H. fringillae. This observation is consistent with a previous study identifying the same parasite genera in this species (Greiner et al. Reference Greiner, Bennett, White and Coombs1975). The ornithophilic vectors that transmit these parasite types include blackflies, Simulium spp. and biting midges, Culicoides spp. (Noblet et al. Reference Noblet, Adkins and Kissam1972; Yu et al. Reference Yu, Wang and Yeh2000; Steele and Noblet, Reference Steele and Noblet2001; Valkiunas and Iezhova, Reference Valkiunas and Iezhova2004; Votypka and Svobodova, Reference Votypka and Svobodova2004; Mullens et al. Reference Mullens, Cardona, McClellan, Szijj and Owen2006; Hellgren et al. Reference Hellgren, Bensch and Malmqvist2008), but no study has identified the specific vectors that transmit blood parasites to crossbills. Thus, it is not known whether associations between blood parasite species (H. fringillae and T. avium) or lack thereof (e.g. H. fringillae and L. fringillinarum) as observed in this study reflect aspects of the vector population dynamics or of the hosts' physiology or behaviour.
Some hatch-year crossbills were infected with L. fringillinarum and/or T. avium, demonstrating that these parasites can be transmitted to young (i.e. a few months old) birds. In adults, the prevalence of the 3 protozoan species changed seasonally in parallel with changes in ambient temperatures. Specifically, the prevalence of L. fringillinarum increased from 0% (January, average temperature below –20°C) to 90% (males in May and June, average temperature approx. 10°C). An increase in L. fringillinarum prevalence was noted as early as March (males: 18%) and April (females: 42%; Fig. 1B). The pre-patent period of Leucocytozoon sp. is thought to be 1–3 weeks (Kocan and Clark, Reference Kocan and Clark1966; Ahmed and Mohammed, Reference Ahmed and Mohammed1978). If the same applies to crossbills and had the late winter increase in L. fringillinarum prevalence been caused by new inoculations, these inoculations would have taken place as ambient temperatures were consistently well below freezing. For example, the average temperature at the study site in mid-March was −12·8°C. Such temperatures are generally considered to be incompatible with ornithophilic vector activity, suggesting that the late winter increase in L. fringillinarum prevalence resulted from relapse and not new inoculations. Consistent with this suggestion, birds can retain infections with Leucocytozoon spp. for at least 1 year after inoculation (Dick, Reference Dick1978; Fallis et al. Reference Fallis, Desser and Khan1974). Previous research led to the proposal that a relapse may contribute to increased blood parasitaemia in other free-ranging birds (Haemoproteus danilewsky in Blue Jays, Cyanocitta cristatamay: Garvin and Schoech, 1996; Leucocytozoon spp. in Redstarts, Phoenicurus phoenicurus: Rintamaki et al. Reference Rintamaki, Huhta, Jokimaki and Squires-Parsons1999), but the present work may provide the best evidence to date in support of this hypothesis.
As was the case for L. fringillinarum, the prevalence of H. fringillae and T. avium was nil to low during winter. Contrary to the situation for L. fringillinarum, however, the prevalence of these species did not increase until mid-May. Ambient temperatures at this time were above freezing and gradually increasing, and this was presumably associated with emergence of vectors (Werner, Reference Werner1992). The degree of haemoparasitic infections in other avian species increases as a function of the vector abundance (Bennett et al. Reference Bennett, Thommes, Blancou and Artois1982; Godfrey et al. Reference Godfrey, Pence and Fedynich1990; Merilä et al. Reference Merilä, Bjorkund and Bennett1995). These observations suggest that new inoculations contributed to the spring increase in prevalence of H. fringillae and T. avium seen in adult crossbills. Due to the absence of data on the identity of ornithophilic vectors in the study area and the phenology of vector emergence we can, however, not conclusively dissociate effects on H. fringillae and T. avium prevalence due to new inoculations from those potentially resulting from relapse (in the case of H. fringillae).
Control of late winter relapse of L. fringillinarum
What accounts for the putative late winter relapse of L. fringillinarum seen in crossbills? Based on previous work demonstrating correlations between blood parasitaemia and T and showing effects of this hormone on haematozoan infections (Introduction section), a first possibility is that T, which in male crossbills was seasonally elevated in late winter, stimulated a relapse of L. fringillinarum. Plasma T has to our knowledge not been measured in female crossbills but in other passerines, females during the breeding season generally have substantially lower plasma T than conspecific males (Introduction section). Considering all the adult samples, male crossbills had on average higher blood prevalence of L. fringillinarum than female crossbills. However, this was not always the case. In particular, L. fringillinarum prevalence was not sexually dimorphic in March and April, around the time that plasma T was elevated in males. Furthermore, the prevalence of L. fringillinarum was higher in males than females in late May and early June, when plasma T in males had decreased to baseline levels. Finally, there was no statistical correlation in males between plasma T and their prevalence of any parasite species. Collectively, these observations do not support the hypothesis that elevated plasma T caused a late winter relapse of L. fringillinarum. However, firm conclusions on this topic require additional studies examining whether reproductive hormones influence haematozoan infections in females and measuring blood L. fringillinarum prevalence as a function of time during which plasma T is elevated.
A second possibility is that breeding effort, independent of reproductive hormones, induced a relapse of L. fringillinarum. Hatch-year crossbills at the study site in 1999 were first caught in April, and by May these birds comprised 60% of all captures (Deviche and Sharp, Reference Deviche and Sharp2001). Thus, crossbills bred successfully in the study area starting in winter and continuing in spring, at the same time that their blood L. fringillinarum prevalence increased. Previous passerine studies found a trade-off between reproductive investment and resistance to haemoparasitic infections, and the blood H/L ratio, which increases in response to an increase in breeding effort, is positively related to stress (Introduction section). This effort may, therefore, result in physiological stress that in turn causes a relapse of haematozoan infections during the breeding period. However, the present study provides no evidence to this effect: The blood abundance of heterophils and lymphocytes in adult crossbills increased between winter and summer, but H/L ratios did not change seasonally and, in addition, birds were heavier and in better body condition in winter than in summer. These observations do not support the hypothesis that crossbills experienced more stress during (winter) than at the end (summer) of their reproductive season and, therefore, that the putative late winter relapse of L. fringillinarum infection was stress- or breeding effort-related. Studies measuring additional stress-related parameters such as plasma corticosterone are warranted to test this conclusion. Furthermore, it is possible that only some winter-sampled crossbills in this study were actually breeding. If so, an effect of breeding effort on H/L ratios may have escaped detection due to the inclusion of non-breeding birds in the data sets.
Finally, exposure to increasing photoperiod, rather than breeding effort per se, may have stimulated a relapse of L. fringillinarum. Day length in the study area rapidly increases at the end of winter and during spring (Deviche and Sharp, Reference Deviche and Sharp2001) and in the Blackcap, Sylvia atricapilla, transfer from a 12 h- to a 16 h-long day length increased the blood prevalence of Haemoproteus belopolskyi and Trypanosoma spp. even though birds were protected from vectors and did not breed (Valkiunas et al. Reference Valkiunas, Bairlein, Iezhova and Dolnik2004). The mechanism mediating this effect of day length is not identified. The same may occur in crossbills but as the blood of Blackcaps did not contain Leucocytozoon spp., it is unknown whether exposure to long days increases the patency also of this parasite.
Sex differences in blood parasite prevalence
Sex differences in avian haematozoan infections were reported in some previous studies (Red-winged Blackbird, Agelaius phoeniceus: Weatherhead and Bennett, Reference Weatherhead and Bennett1991; Brown-headed Cowbird, Molothrus ater: Weatherhead and Bennett, Reference Weatherhead and Bennett1992), but not in others (Norris et al. Reference Norris, Anwar and Read1994; Rintamaki et al. Reference Rintamaki, Huhta, Jokimaki and Squires-Parsons1999; Deviche et al. Reference Deviche, Greiner and Manteca2001; Hartup et al. Reference Hartup, Oberc, Stott-Messick, Davis and Swarthout2008).
One factor that may contribute to differences between studies is that sex differences in haematozoan prevalence can be age- and parasite species-specific. Illustrating this observation, SY female crossbills had lower H. fringillae prevalence than older females. Furthermore, this difference was specific to H. fringillae and was not seen in males. A second factor is that sex differences in haematozoan infections can be seasonal. For example, adult male crossbills had higher L. fringillinarum prevalence than adult females (91% vs 39%) in late May and early June, but not at other times of the year.
What explains this season-dependent sex dimorphism in haematozoan infections? One hypothesis is that the sex difference in blood parasite prevalence results from reproductive hormone-mediated differences in physiological resistance to infections (Weatherhead and Bennett, Reference Weatherhead and Bennett1991) or in behaviour. Testosterone generally circulates at higher concentrations in males than females and is immunosuppressive (Do Prado et al. Reference Do Prado, Leal, Anselmo-Franci, de Andrade and Kloetzel1998; de Souza et al. Reference de Souza, Rivera, raujo-Jorge and de Castro2001; Deviche and Cortez, Reference Deviche and Cortez2005). As a result, males during summer may be physiologically less resistant than females to the effects of new inoculations on their haematozoan prevalence. Consistent with this view, male rodents are often more susceptible to infectious agents than females (nematode, Haemonchus contortus, in sheep: Gauly et al. 2006; Trypanosoma evansi in rat: Al-Mohammed, Reference Al-Mohammed2006; Plasmodium chabaudi in House Mouse: Cernetich et al. Reference Cernetich, Garver, Jedlicka, Klein, Kumar, Scott and Klein2006; Trypanosoma cruzi in House Mouse: Schuster and Schaub, Reference Schuster and Schaub2001). Alternatively, T may influence haematozoan infections through behavioural changes. Testosterone promotes the propensity of birds to express male-typical behaviours, such as singing and territory defence (Van Duyse et al. Reference Van Duyse, Pinxten and Eens2002; Day et al. Reference Day, McBroom and Schlinger2006; Strand et al. Reference Strand, Ross, Weiss and Deviche2008), that presumably increase exposure to ornithophilic vectors and, therefore, the probability of new infections.
The present data do not exclude the possibility that male crossbills had higher prevalence of L. fringillinarum than females in summer due to a T-mediated sex difference in resistance to new infections or behaviour. However, the sex difference in L. fringillinarum prevalence was noted several months after plasma T in males had decreased to low levels. This finding does not support a short-term role for T in the control of the observed sex difference in L. fringillinarum, but males and females may have differed immunologically or behaviourally during summer due to factors other than their reproductive hormones.
Alternatively, sex differences in plumage brightness, rather than in immune function or behaviour, may have caused male crossbills to be more susceptible to attacks by L. fringillinarum vectors than females. Adult male crossbills generally have a bright red plumage whereas females have a drab green or yellowish plumage (Benkman, Reference Benkman, Poole and Gill1992; personal observation), and interspecific comparisons found a positive relation between plumage brightness and haemoparasitic infections (Hamilton and Zuk, Reference Hamilton and Zuk1982; Scheuerlein and Ricklefs, Reference Scheuerlein and Ricklefs2004). The factors that account for this relation, its generality, and whether it applies intraspecifically, remain debated. For example, Garvin and Remsen (Reference Garvin and Remsen1997) proposed that the reported association between plumage brightness and susceptibility to attacks by vectors reflects the fact that bird species living in canopies are, on average, more colourful than species living closer to the ground and some ornithophilic vectors are most common in tree canopies. Furthermore, a comparative study quantifying plumage colourfulness in Aimophila sparrows found no association between this parameter and blood parasite prevalence (Deviche et al. Reference Deviche, McGraw and Greiner2005) and intraspecifically, male plumage redness was independent of haemoparasite prevalence (Redpoll, Carduelis flammea: Seutin, Reference Seutin1994). Experimental studies examining the preference of vectors for specific plumage colours within avian species are warranted to test the hypothesis that these colours influence the vectors' choice of host.
