Immune response to an endotoxin challenge involves multiple immune parameters and is consistent among the annual-cycle stages of a free-living temperate zone bird

A central premise in ecological immunology is that animals trade off investment into immune function against other competing physiological and behavioural processes (Sheldon and Verhulst, 1996; Lochmiller and Deerenberg, 2000; Norris and Evans, 2000). However, one important problem that ecoimmunologists face is how to distinguish a reallocation of resources away from the immune system from a reallocation or redistribution within the immune system. Reductions in one or more elements of the immune system do not necessarily equate to a net reduction in immune function because other parts of the immune system might be boosted simultaneously (Adamo, 2004). Simultaneous measurements of multiple immune indices can help address this problem (Adamo, 2004; Matson et al., 2006; Boughton et al., 2011; Buehler et al., 2011; Demas et al., 2011). Yet understanding trade-offs and interactions within the immune system requires an experimental challenge of the immune system and subsequent quantification of the response using multiple indices (Martin et al., 2006; Martin et al., 2008; Boughton et al., 2011; Pedersen and Babayan, 2011).

The immune system can be experimentally challenged by injection of an endotoxin such as lipopolysaccharide (LPS), part of the cell wall of gram-negative bacteria (Owen-Ashley and Wingfield, 2007). As gram-negative bacteria are universal in most environments, an experimental challenge with LPS mimics a functionally relevant natural situation. Injection of LPS initiates an immune response by mimicking the first stages of a bacterial infection without actually resulting in sustained disease. This innate response begins minutes after endotoxin detection and defends against threats that breach physical barriers such as the skin. Most experimental studies on induced immune responses in free-living birds so far have focused on hormonal, behavioural or metabolic changes (Bonneaud et al., 2003; Owen-Ashley and Wingfield, 2006; Owen-Ashley and Wingfield, 2007; Owen-Ashley et al., 2006; Adelman et al., 2010; Hegemann et al., 2012b; reviewed by Hasselquist and Nilsson, 2012). Studies in free-living birds that characterize multiple immunological responses and subsystems simultaneously are lacking so far.

In addition to quantifying which parts of the immune system are affected by a simulated infection, experimental immune challenges can also be used to investigate the consistency of responses through time. Immune responses may be constant among annual-cycle stages, or responses may be seasonally reorganized as a result of trade-offs with other physiological and behavioural demands. Hypotheses relate increased energy demands and decreased resource availability to compromises in costly immune functions and shifts towards less costly immune components (Nelson and Demas, 1996; Nelson, 2004; Hasselquist, 2007; Martin et al., 2008). Immunological mechanisms aimed at avoiding autoimmunity (Råberg et al., 1998) and preventing oxidative stress (Sorci and Faivre, 2009) might further influence this process. Several studies on non-induced (baseline) immune function indeed show that different indices express different seasonal patterns among and within annual-cycle stages (Nelson and Demas, 1996; Buehler et al., 2008b; Pap et al., 2010a; Pap et al., 2010b; Hegemann et al., 2012c). Thus, reorganization of baseline immune function appears to depend on both environmental conditions and competing biotic processes. Data on seasonal variation in induced immune responses are scarce. However, these are the data needed to verify the hypothesis that free-living birds switch from costly inflammatory responses to highly specific but less costly antibody responses during demanding times (Lee, 2006). A study on captive red knots (Calidris canutus) provides evidence for saved costs on inflammatory responses during demanding times (Buehler et al., 2009). In contrast, wild skylarks (Alauda arvensis) do not modulate the energetic investment in the acute phase response despite seasonal variation in energetic constraints. They maintain a similar response throughout the annual cycle as measured by metabolic rate, body temperature, body mass loss, ketone and glucose concentrations (Hegemann et al., 2012b). The detailed knowledge of non-induced (baseline) immune function and energetic costs of an immune challenge in skylarks make this species an ideal candidate for studying the response of multiple immune indices during an immune challenge in different annual-cycle stages. Such a study will also provide a way to test whether induced responses are modulated among annual-cycle stages (following patterns of baseline immune function), or whether they are maintained throughout the year (reflecting patterns of energetic costs).

Variability in baseline immunological values might also represent important constraints for responses because the ability to mount an immune response might depend on baseline values. For example, baseline haptoglobin concentrations in pigeons (Columba livia domestica) have some capacity to predict post-challenge response concentrations (Matson et al., 2012a). Great tits (Parus major) with high pre-immunization heterophil/lymphocyte ratios (H/L ratios) mount weaker antibody responses (Krams et al., 2012). However, it remains to be tested in free-living birds if particularly high (or low) baseline values of a given immune parameter limit the responsiveness of that parameter to an immunological stimulus (i.e. ‘immunological ceiling’). In other words, do individuals with relatively high baseline values respond differently to an immune challenge than birds with relatively low levels? The existence of immunological ceilings can have important implications for the interpretation of values collected from field samples.

In this study, we challenged wild skylarks with LPS and compared them with un-injected controls during five annual-cycle stages to test: (1) which immune parameters are affected by an endotoxin challenge; (2) whether the immunological response varies among annual-cycle stages; and (3) whether baseline values present constraints to the magnitude of the immune response. To capture a broad picture of the immune response, we measured different components of immune defence. (1) Natural antibodies and complement, which agglutinate and lyse foreign cells (Matson et al., 2005) and are measures thought to be unaffected by previous exposure (Ochsenbein and Zinkernagel, 2000). (2) The acute phase protein haptoglobin, which limits the role of plasma iron as nutrient for pathogens and is an initiator of oxidative damage (Murata et al., 2004; Quaye, 2008). (3) The relative abundances of leukocytes, which reflect both innate and acquired components of immune function. Leukocytes are circulating continuously through the blood to maintain a state of readiness and are redistributed in response to immunological stimulation (Feldman et al., 2000). Leukocyte analyses include the H/L ratio, which is related to immunological and other stressors (reviewed by Davis et al., 2008). (4) And heat shock proteins (Hsps), which indicate stress (Martinez-Padilla et al., 2004) and have been suggested to be a potential indicator for autoimmune risk (Hasselquist and Nilsson, 2012). Furthermore, they play an important role in modulating innate and acquired immunity (Pockley, 2003; Pockley et al., 2008) through their capacity to activate complement and trigger the release of inflammatory cytokines (Calderwood et al., 2007).

We caught adult skylarks Alauda arvensis Linnaeus 1758 during five annual-cycle stages in the northern Netherlands in 2008, focusing on our study population at the Aekingerzand [52°55′N, 6°18′E (Hegemann et al., 2012b)]. Some skylarks in our study population migrate; others winter locally and are accompanied by birds that breed further north and east (Hegemann et al., 2010). We caught birds during breeding in June and July (nine males, six females), moulting in August and September (12 males, seven females), autumn migration in October (12 males, 12 females), wintering in December and January (14 males, three females) and spring migration in March (17 males, nine females). Birds were sexed biometrically, and in some doubtful cases molecularly (Hegemann et al., 2012a). For details on catching, see Hegemann et al. (Hegemann et al., 2012b). All individuals were fully grown. Because skylarks undergo a complete post-nuptial moult in August–September, age classes could not be distinguished. As skylarks breed in their first year (A.H., unpublished data) and both young and adult birds are known to migrate (van Dobben and Mörzer Bruyna, 1939; Hegemann et al., 2010), we have no indications that an age bias between stages exists and could influence the interpretation of the results.

When catching skylarks in the field we collected blood (~150 μl) into heparinized capillary tubes from the brachial vein after capture (median: 5 min; range: 2.25–30 min) to minimize any impacts of handling stress (Buehler et al., 2008a). We then took structural measurements. We refer to measurements from these samples as ‘field values’.

After capture, birds were brought into captivity (cages 30×40×60 cm). During the breeding season, when skylarks are territorial, birds were housed individually. During the non-breeding seasons, when skylarks live in socially interacting flocks, birds were housed in small groups (≤3 birds per cage). Even though the captivity period was short, we attempted to avoid a potential seasonal bias by using conditions that reflected current social conditions in the wild. Birds had ad libitum access to water and food (mealworms and seeds) until 16:30 h on the experimental-protocol start day (for details, see Hegemann et al., 2012b).

We started the experimental protocol by isolating birds in a dark box for 1 h without food and water. At 17:30 h we injected experimental birds with 2.5 mg LPS in 10 ml phosphate buffered saline (PBS) per kilogram body mass in their abdominal cavities (Hegemann et al., 2012b). Control birds remained un-injected, because puncturing the skin and underlying tissues for injecting only a vehicle (i.e. PBS) can result in inflammation (K. Klasing and B. Helm, personal communications). Consequently, the experimental responses must be viewed as a result of both LPS and injection procedure. This combination does not pose interpretational problems for our study because our central interest is the immune response and not the effects of LPS per se. After injection the experimental birds and their corresponding controls were put into dark boxes (metabolic chambers) where they spent the night under thermo-neutral conditions (Hegemann et al., 2012b). The next morning at 06:30 h (13 h after injecting experimental birds) we collected another blood sample (150 μl) within 10 min of removing birds from boxes. The 13 h interval between the start of the experiment and taking a blood sample was based on the ability to match metabolic measurements with blood sampling in a time frame in which most physiological and behavioural reactions occur (e.g. Owen-Ashley et al., 2006; Adelman et al., 2010; Burness et al., 2010), and with the need to return birds, especially during the breeding season, quickly to the field.

From each blood sample (field and laboratory), we used a small drop to make blood smears for leukocyte enumeration. The remainder of each sample was centrifuged at 2600 g for 10 min. Plasma and red blood cells were separated and stored at −20°C. Upon completion of the protocol, birds were released at the site of capture.

Because the stress of short-term captivity might affect immune function differently in different seasons (Sapolsky et al., 2000; Martin, 2009), we evaluated the effects of stress throughout the annual cycle. For this purpose, we used H/L ratios (Gross and Siegel, 1983; Vleck et al., 2000; Davis, 2005; Huff et al., 2007) and Hsp70 concentrations (Martinez-Padilla et al., 2004; Bourgeon et al., 2006). We favoured these two independent and functionally integrative methods over concentrations of specific hormones (e.g. corticosterone) because the effects of hormones can depend strongly on levels of binding globulins and other related factors (Deviche et al., 2001; Lynn et al., 2003). In order to separate the stress of captivity and general protocols (experienced by all birds) from the stress response of the immune challenge (experienced only by experimental birds), we explored seasonal variation in stress response in the control birds. With these birds, we calculated differences between values from field samples and morning laboratory samples (ΔH/L ratio and ΔHsp70). While both indices increased as expected due to the stress associated with captivity, we found no significant seasonal pattern in this captivity-related stress response (ΔH/L ratio χ²_4,46=4.04, P=0.40; ΔHsp70 χ²_4,45=0.40, P=0.53). Furthermore, there were no differences in the metabolic effects (O₂ consumption and nightly mass loss) of an LPS injection when comparing birds that were held in captivity for a few hours according to the protocol used in this study and birds acclimated to captivity for 55 days (Hegemann et al., 2012b). Thus we have no evidence that the immune response we experimentally triggered was masked by any stress responses resulting from the short time in captivity. Experiments were performed under license DEC5219B of the Institutional Animal Care and Use Committee of the University of Groningen.

We used a hemolysis-hemagglutination assay (rabbit red blood cells, B-0009H; Harlan Laboratories, Leicestershire, UK) to quantify titres of complement-like lytic enzymes (i.e. lysis) and non-specific natural antibodies (i.e. agglutination) in plasma (Matson et al., 2005; Hegemann et al., 2012c). Scans of individual samples were randomized among all plates and scored blindly to treatment and season (by A.H.). A plasma standard was run in duplicate in all plates. On average, variation (standard deviation) within (0.4 lysis titres and 0.7 agglutination titres) and among (0.5 lysis titres and 1.1 agglutination titres) plates is similar to that originally described by Matson et al. (Matson et al., 2005). We used a commercially available colorimetric assay kit (TP801, Tri-Delta Diagnostics, Morris Plains, NJ, USA) to quantify haptoglobin concentrations (mg ml⁻¹) in plasma samples (Hegemann et al., 2012c; Matson et al., 2012b). Blood smears were examined by one person (C. Gottland), who was blind to treatment and season. The first 100 white blood cells (WBCs) per slide were identified and counted as lymphocytes, heterophils, basophils, monocytes or eosinophils (Hegemann et al., 2012c).

Cell lysates were obtained as in Tomás et al. (Tomás et al., 2004) and total protein concentration was determined by the Bradford method using bovine serum albumin as the standard. Concentrations of Hsp70 were determined from the cell lysates by means of an enzyme-linked immunosorbent assay using the protocol described by Mahmoud and Edens (Mahmoud and Edens, 2003). Briefly, 100 μl of samples (dilution 1:10), standards (0–50 ng recombinant human Hsp70) and a positive control (HeLa cell lysate) were coated in duplicate in 96-well immunoplates at 4°C overnight. After blocking non-specific binding sites, plates were incubated for 1 h with 100 μl of anti-Hsp70 monoclonal antibody (H5147, Sigma-Aldrich, Madrid, Spain) diluted 1:1000. Following washing, plates were incubated with 100 μl of 1:5000 alkaline phosphatase conjugated goat anti-mouse IgG polyclonal antibody (SAB-101; Enzo Life Sciences, Lausen, Switzerland) for 1 h. Finally, we added 1 mg ml⁻¹ p-nitrophenylphosphate in coating buffer for 30 min, and read the absorbance of individual wells at 405 nm with a microplate reader (PowerWave, BioTek, Bad Friedrichshall, Germany). Hsp70 concentration was calculated from the standard curve. All final Hsp70 values were standardized by dividing Hsp70 concentration by total protein and normalized according to plate-specific positive controls to facilitate inter-plate comparisons. Based on samples run in duplicate, the mean intra-assay coefficient of variation was 5.9% and the mean inter-assay coefficient of variation was 6.6%.

We compared experimental and control groups for each response variable using linear models analysed in R version 2.9.2 (R Development Core Team, 2009). We included treatment, annual-cycle stage, sex and all possible interactions as explanatory variables. WBC types were analysed with generalized linear models with a quasi-binomial approach and F-tests. These tests incorporated the counts of one cell type and the total remaining WBC number (e.g. basophils against the sum of heterophils, lymphocytes, monocytes and eosinophils using the ‘c-bind’ function in R). H/L ratios were tested in a linear model.

To test whether baseline values as taken upon capture in the field affected the outcome of the experiment, we calculated the individual deviation from season- and sex-specific means. We included this sex- and season-independent term and the interaction with treatment in all analyses. A ceiling in the ability to respond would be indicated by a significant interaction. A significant main effect would indicate that individuals express consistent immune parameters in the field and in the laboratory after having gone through a standardized experimental protocol in the preceding 14 h period.

We always started with the full model and simplified it using backwards elimination based on log likelihood ratio test with P<0.05 as the selection criterion (‘drop1’ in R) until reaching the minimal adequate model. Model assumptions were checked using the residuals of the final model. Sample sizes differ among response variables because of insufficient plasma volume. Graphs were made using the package ‘gplots’ (Warnes, 2009). Baseline values, measured in the field just after capture, did not differ significantly between experimental and control groups in any of the 10 parameters measured (always P>0.25).

Compared with control birds, injected skylarks exhibited significantly increased lysis 13 h after an endotoxin challenge (Fig. 1A), but experimental and control birds did not differ significantly in terms of agglutination (Fig. 1B). Concentrations of haptoglobin were significantly lower in endotoxin-challenged birds (Fig. 1C). Experimental birds had significantly higher proportions of heterophils than control birds (Fig. 1E). The proportions of lymphocytes, basophils and eosinophils were lower in experimental birds (Fig. 1F,B,I). The H/L ratio, the proportion of monocytes and concentrations of Hsp70 were not affected by the endotoxin challenge (Fig. 1D,H,J). Thus, experimental birds differed significantly from control birds in six of the 10 physiological parameters (Table 1). We never found a significant difference between the sexes in their response to the endotoxin challenge (treatment×sex interaction, all χ²/F<1.27, P>0.26). Independent of treatment, males and females differed significantly in two of the 10 parameters in the morning after the injection (Table 1). Females exhibited significantly higher proportions of eosinophils among their WBCs (females 10.6%, males 5.9%). Males had significantly higher haptoglobin concentrations (15.9%) and statistically marginally higher lysis titres (16.1%) than females.

Based on samples taken in the morning in the laboratory, Skylarks showed significant differences among annual-cycle stages in eight of the 10 measured parameters, but the immune response after the endotoxin challenge did not differ among annual-cycle stages: the interaction between annual-cycle stage and endotoxin challenge was not significant for any of the measured parameters (Table 1, Fig. 1).

The response to the endotoxin challenge was independent of field immune values. Changes in immune parameters after the endotoxin challenge were always independent of the corresponding value measured in the field (treatment×field value deviation interaction, all χ²/F<3.12, P>0.08).

After accounting for treatment, individual skylarks showed values that were consistent between deviation of the field values and morning samples for five parameters. With lysis titre, haptoglobin concentration, the proportion of heterophils, lymphocytes and eosinophils, we found significant positive relationships between the field values (corrected for sex and season variation) and the morning laboratory values (Table 1). There was no significant relationship between deviation of the field values and morning values for agglutination titre, Hsp70 concentration, H/L ratio or the proportion of basophils and monocytes (Table 1).

Skylarks exhibited complex and multifaceted responses when experimentally challenged with endotoxin. Thirteen hours post-injection, some parameters increased (lysis titre, heterophil proportion), others decreased (haptoglobin concentration, proportion of lymphocytes, basophils and eosniophils) and others were unchanged (agglutination titres, H/L ratio, Hsp70 concentration and proportion of monocytes). The complexity of the immune response to endotoxin highlights methodological complications for ecoimmunologists trying to interpret samples and data collected from birds in the field. For example, relatively high or low values of one immune parameter should be interpreted cautiously when measured in isolation of other parameters. Despite its complexity, we found no evidence for seasonal reorganization of the immune response, which was consistent among five annual-cycle stages. Furthermore, we found no evidence for an immunological ceiling; birds showed similar responses to an immunological challenge regardless of their baseline values measured from values collected in the field. After accounting for treatment, individuals showed consistent values in samples from the field and samples from the laboratory for lysis titre, haptoglobin concentration, H/L ratio and eosinophil proportions. This suggests that these measures are relatively robust against possible sources of variation such as time of day, activity and nutritional status.

One particularly surprising result that highlights the complications associated with assigning a single immune parameter relates to haptoglobin. LPS-injected skylarks exhibited significantly lower concentrations of haptoglobin 13 h post-challenge compared with un-injected control birds. Haptoglobin is an acute phase protein that is released from the liver during a pathogenic challenge. Normally in birds, concentrations of haptoglobin or iron-binding functional equivalents increase in association with inflammation (Thomas, 2000; van de Crommenacker et al., 2010). Our finding suggests that in skylarks haptoglobin might be more appropriately classified as a negative, rather than a positive, acute phase protein when measured 13 h after the endotoxin challenge. Any functional relevance of the observed reductions in concentrations of this protein, which sequesters iron, remains to be elucidated. Notably, compared with other species that have been similarly assayed, skylarks maintain relatively high circulating concentrations of baseline haptoglobin (Matson, 2006). The decrease we observed in skylarks following LPS injection may relate to dissimilar rates of haptoglobin production and consumption in this species. This result also suggests a greater reliance of skylarks on constitutive (rather than induced) production of this bacteriostatic and antioxidant molecule. Testing these possibilities will require more detailed studies (e.g. with more frequent sample time points) of the LPS-induced inflammation time course in skylarks and other species of birds. However, results of a pilot study showed that haptoglobin concentrations in skylarks decreased by 13 h and remained low at 24 h after an LPS injection (K.D.M., unpublished data).

Lysis titres of skylarks increased following endotoxin challenge, but there was no difference in agglutination (natural antibody) titres between control and experimental birds. Antibody production normally requires days, not hours. Thus it is not surprising that agglutination titres did not differ between groups 13 h post challenge. The increase in lysis titres during infection points to another important topic of ecoimmunology: high values are not necessarily better (De Coster et al., 2011). Instead, values should be viewed in relation to the immunological status, e.g. by measuring parasite infection rates (Pap et al., 2011).

Circulating leukocytes are important for the protection against invading microorganisms. During immune responses, redistributions of leukocyte populations occur (Gehad et al., 2002). In skylarks, proportions of heterophils increased and proportions of lymphocytes decreased following endotoxin challenge. Because heterophils relate to innate immunity and lymphocytes relate to acquired immunity, the innate inflammatory response we elicited could primarily affect heterophil concentrations (De Boever et al., 2009). However, upon an immune challenge, a redistribution of peripheral blood lymphocytes to secondary lymphoid organs occurs (Gehad et al., 2002), and this process could contribute to reduced numbers of lymphocytes in the circulating blood. Basophils are one of the first leukocyte types to enter tissue during an early inflammatory response in birds (Katiyar et al., 1992). The decreased proportion of basophils suggests that these cells are no longer circulating in the peripheral blood and have migrated into the tissue at the LPS injection site. Given this biological functioning, we are confident that all observed changes in leukocyte profiles are meaningful, despite two relatively high P-values, which should be viewed with added caution in multiple testing (Moran, 2003).

We found no difference in concentrations of intracellular Hsp70 concentrations between control and experimental skylarks. Autoimmune reactions caused by physiological stress during an immune response might be an important cost of immunity (Råberg et al., 1998) and heat shock protein quantification could be an indirect way to assess the potential risks of autoimmune reactions (Hasselquist and Nilsson, 2012). Our data do not provide any evidence for increased physiological stress during the immune response to an endotoxin. Heat shock proteins also have more direct immunological functions. Extracellular levels of certain types, such as Hsp60 and Hsp70, exhibit modulating effects on innate and acquired immunity (Pockley, 2003; Pockley et al., 2008), and these proteins are involved in the activation of complement and release of cytokines (Calderwood et al., 2007). In rats, intracellular and extracellular heat shock protein concentrations are correlated (Fleshner et al., 2004). The lack of change in intracellular heat shock protein concentrations in immunologically challenged skylarks, despite increased complement activity, might indicate different relationships in wild birds. Detailed studies of both extracellular and intracellular heat shock protein concentrations at multiple time points following an immunological challenge are required to reveal the causes and consequences of heat shock protein variation in wild birds.

We found no evidence that the response of the immune system to endotoxin differed among five annual-cycle stages experienced by skylarks. The reaction to the endotoxin challenge also did not differ between the sexes. These findings are in line with our previous finding that energetic components of an acute phase response (as measured by metabolic rate, body temperature, body mass loss, ketone and glucose concentrations) are not seasonally modulated in this species (Hegemann et al., 2012b). After statistical correction of the treatment effects, it is noteworthy that we found seasonal differences in eight of 10 immunological parameters that we measured using samples collected in the laboratory. These data support our earlier findings that free-living skylarks modulate their baseline immune function among annual-cycle stages as measured on samples collected upon capture in the field (Hegemann et al., 2012c). Taken together, our results suggest that skylarks do modulate baseline values of immune function, as has been described for other species (Buehler et al., 2008b; Pap et al., 2010a; Pap et al., 2010b). However, both the energetic (Hegemann et al., 2012b) and the immunological (present study) consequences of an endotoxin challenge are constant throughout the year, independent of other annual-cycle demands and equal for both sexes. This suggests that mounting this type of immune response is crucial to survival and cannot be compromised. Only baseline defences can be traded off with other demands. This conclusion further highlights the interpretational limitations and the importance of distinguishing between baseline values and induced responses when studying ecological immunology (Adamo, 2004; Hegemann et al., 2012b; Hegemann et al., 2012c).

This finding – that responses to an LPS injection were constant throughout the annual cycle – necessitates a short discussion of two methodological points. First, skylarks in our study population are partial migrants: some birds migrate; others winter locally and are accompanied by birds from more northern and eastern breeding populations (Hegemann et al., 2010). With our year-round study focused on the breeding location, we potentially caught a mixture of birds from different populations during wintering and migration. However, similar coefficients of variation throughout the annual cycle for each of the response variables demonstrate that any (unmeasured) variability in the composition of the sampled populations did not translate to differences in immunological variability. This is also supported by our data on baseline immune function (Hegemann et al., 2012c) and energetic effects of an immune challenge (Hegemann et al., 2012b). Consequently, immune responses to LPS injection by skylarks seem to be relatively independent of breeding location and more dependent on current local conditions. Second, as with any decision based on statistics, our acceptance of our null hypothesis was influenced by sample size, variance and effect size. Small sizes can undermine some hypotheses via type II errors or false negatives, but this is unlikely to occur consistently (e.g. as with all 10 dependent variables). Large sample sizes can minimize this type of error but sometimes detect differences that lack biological relevance. In principle, power analyses could provide insight into these issues, but in practice such analyses require precise input regarding strength, direction, timing and variance, none of which are available for the interaction between LPS treatment and annual-cycle stage. In light of these points, we feel confident that our sample sizes are sufficient to accept our null hypothesis and draw conclusions accordingly.

The strength of the immunological response as measured by 10 parameters was independent of the corresponding values measured upon capture in the field. Thus birds did not face an immunological ceiling, and similar immune responses were mounted regardless of an individual's baseline values. A corresponding pattern also exists at the population level as the immune response was independent of seasonal patterns of baseline values (see above).

Discarding the effect of the endotoxin challenge, several immune indices (lysis titre, haptoglobin concentration, and the proportion of heterophils, lymphocytes and eosinophils) show a significant correlation between values from samples collected in the field and values from samples collected in the laboratory after birds had gone through a standard 14 h protocol. After correction for treatment these parameters showed a positive correlation between the field and morning values. These results indicate that individuals exhibit consistent values in the face of variable environmental and physiological conditions. Birds sampled in the morning in the laboratory exhibited highly standardized conditions (temperature and light regime, food and water availability). However, in the field, at least some conditions varied, such as time of day and previous activity. While many factors are known to affect immune indices [e.g. diurnal patterns (Navarro et al., 2003; Martinez-Padilla, 2006); flight behaviour (Matson et al., 2012b)], skylarks showed consistent values for these five parameters independent of the conditions under which they were taken. This suggests that these indices are robust against short-term (hours) biotic and abiotic environmental variation. Consequently, they are suitable for ecoimmunologists interested in longer-term environmental variation or in the immunological status of their study subjects.

Many volunteers helped catch skylarks at various times of the year. We are thankful to everyone, especially Kees van Eerde, Rob Voesten, Martin Keiser, Richard Ubels and Chris Trierweiler. Cecile Gottland and Emmanuelle Gilot performed the slide counts. Isabel Piedad provided assistance in Hsp analysis. Ido Pen gave valuable advice on statistical tests. Discussions with Christiaan Both and his comments on early drafts were of great help. Debbie M. Buehler, Lusia Mendes and three anonymous reviewers commented on an earlier version of the manuscript.