Different incubation patterns affect selective antimicrobial properties of the egg interior: experimental evidence from eggs of precocial and altricial birds

Microorganisms are considered as important selective agents affecting the evolution of life-history traits in different animal taxa (McFall-Ngai, 2002; Evans et al., 2017; Hird, 2017). Avian reproductive strategies and overall reproductive success are probably under the strong selective pressure of environmental microorganisms. Recent studies have found bacterial density on the eggshell to be the main precursor of hatching failure across avian species (Lee et al., 2017; Peralta-Sanchez et al., 2018; Tomas et al., 2018). These findings are not surprising given that previous studies documented that microbes are able to pass through the eggshell into the egg interior and negatively affect embryo viability (Bruce and Drysdale, 1994; Pinowski et al., 1994; Cook et al., 2003; Fonseca et al., 2011; Hansen et al., 2015) or hatchling phenotype (Javurkova et al., 2014).

Eggshell microbial assemblages are determined by substrates that are in close contact with laid eggs, i.e. nest material and skin or feathers of the incubating parent (Martinez-Garcia et al., 2016; Ruiz-Castellano et al., 2016; van Veelen et al., 2018). Additionally, environmental factors such as latitude and humidity were found to affect eggshell microbial assemblages (Cook et al., 2003, 2005b; Wang et al., 2011). Considering the high diversity of eggshell microorganisms (Grizard et al., 2014; van Veelen et al., 2018), including pathogens that are able to invade egg contents (Wang et al., 2018), birds have evolved many defence mechanisms to minimize the risk of microbial trans-shell infection (D'Alba and Shawkey, 2015). Among them, variation in the eggshell microstructure (D'Alba et al., 2014; Grellet-Tinner et al., 2017), eggshell pigmentation (Ishikawa et al., 2010), deposition of egg white antimicrobial proteins (AMPs) into the eggshell membranes and cuticle (Bain et al., 2013; Gautron et al., 2011; Wellman-Labadie et al., 2008a), and cuticle nanostructuring (D'Alba et al., 2016, 2017) have been documented to significantly reduce the eggshell microbial load. Among behavioural mechanisms, partial and full incubation of a clutch have been shown to significantly reduce the abundance (Bollinger et al., 2018; Cook et al., 2005a; D'Alba et al., 2010a; Ruiz-De-Castaneda et al., 2011, 2012) but also the diversity and community structure of the eggshell microbiota (Grizard et al., 2014, 2015; Lee et al., 2014). In contrast, defence mechanisms in species of the family Megapodiidae, which do not incubate their clutches, lie in the covering of the eggshells with an inorganic layer of nanometer-scale spheres of hydroxyapatite, making the eggshells super-hydrophobic and preventing attachment by bacteria (D'Alba et al., 2014).

However, microbes may enter the egg interior despite such sophisticated external defence systems. Thus, a second line of defence in the form of egg white AMPs exists, protecting the developing embryo against invading microbes. Recent studies have identified more than 180 different egg white proteins in chicken eggs (Mann and Mann, 2011; Sun et al., 2017). Of these, lysozyme and ovotransferrin are the most abundant (Burley and Vadehra, 1989), having physiological (Giansanti et al., 2012; Javurkova et al., 2015; Krkavcová et al., 2018) and strong antimicrobial potential (Baron and Jan, 2011; Wellman-Labadie et al., 2008b). While lysozyme has significant bactericidal activity against Gram-positive as well as Gram-negative bacteria (Ibrahim et al., 2001; Wellman-Labadie et al., 2008a), ovotransferrin is primarily bacteriostatic via its iron-binding potential, forming an iron-deficient environment, inhibiting the growth of many microorganisms (Wu and Acero-Lopez, 2012). However, the concentration and antimicrobial activity of egg white lysozyme and ovotransferrin have been documented to vary during the incubation cycle (Fang et al., 2012a,b; Qiu et al., 2012; Grizard et al., 2015; Liu et al., 2015). These changes in AMP activity are considered to be linked with alterations in egg white pH (Smolelis and Hartsell, 1952; Ibrahim et al., 1994; Halbrooks et al., 2005). Egg white pH was observed to vary significantly in chicken eggs during incubation phases (Cunningham, 1974). It seems therefore that the antimicrobial properties of avian egg white are highly dependent on the temperature and/or intensity of incubation.

Birds across the altricial–precocial spectrum, however, vary significantly in incubation behaviour (Deeming, 2002). For altricial birds, the onset of full incubation usually begins with the first laid egg in the clutch, whereas precocial birds only partially incubate the first-laid eggs in a clutch and usually start with full incubation after clutch completion (Loos and Rohwer, 2004; Wang and Beissinger, 2011). It is worth noting that partial incubation has also been observed for some altricial bird species (Wang and Beissinger, 2009). Furthermore, concentrations of egg white AMPs vary significantly across species, with precocial birds tending to have a higher concentration of lysozyme compared with altricial birds (Saino et al., 2007; Shawkey et al., 2008; Bonisoli-Alquati et al., 2010), whereas the concentration of ovotransferrin seems to be less variable between species (Shawkey et al., 2008). It is therefore highly probable that these variabilities in incubation pattern and egg white AMP concentration may significantly interact and have fundamental, yet equivocal effects on antimicrobial potentials of precocial/altricial eggs. Despite this, except for one study (Grizard et al., 2015) investigating egg white AMP concentration during the incubation cycle in the free-living altricial red-capped lark (Calandrella cinerea), all previous studies have only focused on precocial chicken eggs. In addition, another overlooked fact is that the antimicrobial response of egg white to external microbial pressure can be specific to certain microorganisms (Bedrani et al., 2013a,b) that may also include potentially beneficial invaders. The incubation-mediated effect on egg white AMPs may thus favour the proliferation of beneficial microorganisms inside the egg at the expense of pathogens or vice versa. However, to our knowledge, there is no experimental study investigating the role of different incubation patterns, egg white AMP concentrations and their interaction on the antimicrobial properties and components of altricial and precocial eggs regarding potentially pathogenic and beneficial egg invaders.

In this study, we experimentally tested the effect of partial and full incubation, increased egg white lysozyme and ovotransferrin concentrations, and interactions between these two factors on the antimicrobial protection of developing embryos in representative precocial and altricial eggs. Egg white antimicrobial activity was determined against two Gram-positive bacteria, the potentially pathogenic Micrococcus luteus (Pekala et al., 2018) and the beneficial probiotic Bacillus subtilis (Forte et al., 2016a,b; Li et al., 2016; Bai et al., 2017). Partial incubation would have a more pronounced effect on the inhibition of potentially pathogenic bacteria and the maintenance of beneficial probiotic bacteria in precocial eggs. In altricial eggs, we expected a stronger effect of full incubation on the inhibition of pathogenic bacteria and the maintenance of beneficial bacteria. Further, based on previous studies, we expected that egg white ovotransferrin concentration would increase along with the intensity of incubation, whereas lysozyme concentration and pH would decrease.

For this study, we used the eggs of two typical precocial and altricial species, Japanese quail (Coturnix japonica, Temminck and Schlegel 1849) and domestic pigeon (Columba livia, Gmelin, 1789), respectively. Quails were housed at the Institute of Animal Science in Prague-Uhrineves (Czech Republic). Pigeons were kept in open dovecots by two breeders (Petr Smrček and Stanislav Česnek) in Trnava village (Czech Republic). Quails and pigeons were fed ad libitum with a standard feed mix and had free access to water. Our model species strongly differ in their reproductive strategies and incubation patterns. Quails lay relatively large clutches (8–10 eggs), partially incubate their eggs during the egg-laying period and postpone the onset of full incubation until clutch completion (Mills et al., 1997), whereas pigeons lay only two eggs, partial incubation is lacking or minimized and the onset of incubation usually begins with the first laid egg (Vatnick and Foertsch, 1998).

To conduct controlled experiments, we first assessed species-specific differences in total egg mass, egg white volume and concentration of egg white AMPs in the precocial and altricial eggs. Pre-experimental measurements were performed on freshly laid quail eggs (n=18), randomly collected within two consecutive days as the third and fourth eggs in the laying sequence, and freshly laid pigeon eggs (n=18) randomly collected within two consecutive days as the first and second eggs. The eggs were weighed to the nearest 0.01 g using a KERN CM50-C2 mol l⁻¹ digital balance. To obtain egg white volume (mean±s.d.), the eggs were cracked, the egg whites and yolks carefully separated and each egg white weighed to the nearest 0.001 g. Mass was converted to volume (nearest 0.01 ml) based on the measurement of 1 ml of egg white having a mass of 0.963 g for both altricial and precocial egg whites. This method was also used in our previous experimental studies (Javŭrková et al., 2015; Krkavcová et al., 2018) and enabled us to avoid possible inaccuracies in direct volume measurements caused by egg white viscosity. Mean egg white volume for precocial and altricial eggs was 18.0 and 6.8 ml, respectively. These values were then used for calculating the total lysozyme and ovotransferrin content per whole egg in precocial and altricial eggs (see below). Each egg white was then transferred into a plastic vial and immediately analysed for AMP concentration.

Two major egg white AMPs, lysozyme and ovotransferrin, were analysed in (a) pre-experimental precocial and altricial eggs for the purpose of manipulative experiments and (b) experimental eggs after experimental treatments in order to assess post-experimental concentrations of egg white lysozyme and ovotransferrin.

Lysozyme concentration was measured using an agar well diffusion assay (Osserman and Lawlor, 1966). Briefly, Britton–Robinson buffer (pH 7.0) containing 0.492 g boric acid (Alchimica, Prague, Czech Republic), 0.782 g phosphoric acid (98%; Lach-ner, Neratovice, Czech Republic), 0.480 g acetic acid (Lach-ner) and 0.840 g NaOH (Alchimica, Prague, Czech Republic) dissolved in 305 ml of distilled water was prepared. Then, 50 mg of lyophilized Micrococcus lysodeikticus (Sigma-Aldrich, St Louis, MO, USA; ATCC 4698, M3770) was re-suspended in 10 ml of Britton–Robinson buffer and added to a 60°C solution of 1% agar [1 g of agar (Alchimica) re-suspended in 100 ml of Britton–Robinson buffer]. Agar was poured into Petri dishes, left for 30 min to solidify, and 3 mm diameter holes were punched into the agar using a core borer. Each egg white sample was homogenized in a glass vial using a magnetic stirrer (RH Digital, IKA, Oxford, UK) with an immersed cross spin magnetic stirrer bar (12×30 mm) at 1800 rpm for 15 min. Then, 10 μl of properly homogenized and vortexed egg white samples was transferred in duplicate into the holes in the agar plates using pipette tips for viscous liquids (GENO-DNA S tips, CS960 9405120, Thermo-Fisher Scientific Inc., Waltham, MA, USA). Standard solutions (10 µl samples) of known concentration (20, 15, 7, 4, 2, 0.5 mg ml⁻¹) prepared by the dilution of lyophilized hen egg white lysozyme (L6876, Sigma-Aldrich) in Britton–Robinson buffer were added into the holes of each agar plate. The plates were incubated for 24 h at 21°C and 50–60% humidity. Photographs of plates with clearance zones around the holes were taken in a standard position with a Canon camera (EOS450D) using a 50 mm macro objective (F2.8). Diameters of clearance zones were analysed using ImageJ. Lysozyme concentration (mg ml⁻¹) for each egg white sample was interpolated from a calibration curve using GraphPad Prism version 6.00 for Windows (GraphPad Software, San Diego, CA, USA).

Ovotransferrin concentration was measured using a modified version of the iron-binding activity assay (Yamanishi et al., 2002) described in detail by Horrocks et al. (2011). Briefly, 25 μl of properly homogenized and vortexed egg white samples was transferred into a 96-well microplate (BRAND^® microplate, pureGrade, flat-transparent, Sigma-Aldrich) in quadruplicate. To create a calibration curve, a stock solution of ovotransferrin containing 40 mg of ovotransferrin (synonym Conalbumin, C0755, Sigma-Aldrich) dissolved in reagent 1 containing 200 ml ddH₂O, 7.3 g Tris, 6.4 g Na₂CO₃ and 0.84 g Triton-X was prepared; 25 μl of ovotransferrin standard solutions ranging in concentration from 30 to 0.1 mg ml⁻¹ was added in duplicate into the wells of the bottom rows. Then, 120 µl of reagent 2 containing 150 ml of reagent 1 and 600 μl of iron standard solution was added to each well, the plate was shaken for 10 s and ‘pre-reads’ of absorbance at 570 and 660 nm wavelength were measured using a UV/Vis microplate reader (TECAN Infinite1200 PRO, Tecan Group Ltd, Männedorf, Switzerland). After incubation of the plate at 37°C for 5 min, 25 µl of ascorbic acid solution (100 ml ddH₂O, 0.49 g FerroZine™, 0.6 g Tris and 0.574 g ascorbic acid) was added to each well and incubated again (5 min, 37°C). Finally, 100 μl of reagent 3 (200 ml ddH₂O, 25.2 g citric acid and 0.38 g thiourea) was added and absorbance at 570 and 660 nm wavelength was recorded immediately (t=0) and after 6 min of incubation at 37°C (t=7). Absorbance values were corrected for initial values of well-specific ‘pre-read’ absorbance and normalized using values of reference absorbance at 660 nm. The difference between values measured at t=0 and t=7 was used for calculation of ovotransferrin concentration via interpolation of the standard curve in GraphPad Prism version 6.00.

For manipulation experiments, we used freshly laid quail eggs collected within two consecutive days as the third and fourth eggs in the laying sequence and freshly laid pigeon eggs randomly collected within two consecutive days as the first and second eggs. Experimental eggs were produced by 120 breeding quail and 60 pigeon females (i.e. each female produced a maximum of two eggs for our purposes). Eggs were placed into plastic portable boxes and stored in the dark at room temperature for no more than 24 h.

A 4×3 experimental design was followed. In each experimental set, 60 quail and 36 pigeon eggs were marked and randomly assigned into four experimental groups based on AMP treatment: (1) L+: eggs manipulated in ovo with lysozyme; (2) O+: eggs manipulated in ovo with ovotransferrin; (3) L+O+: eggs manipulated in ovo with lysozyme and ovotransferrin; and (4) control: eggs with an in ovo injection of sterile phosphate-buffered saline (PBS). Eggs in these four groups were furthermore divided into another three groups based on incubation treatment: (1) ‘no’: un-incubated eggs; (2) ‘partial’: partially incubated eggs; and (3) ‘full’: fully incubated eggs. For manipulations with AMPs, commercially available lysozyme (L6876, Sigma-Aldrich) and ovotransferrin (C0755, Sigma-Aldrich) purified from hen egg whites were used. Although several types of lysozyme exist, variously distributed across egg whites of avian species (Thammasiririak et al., 2007; Callewaert and Michiels, 2010; Sun et al., 2017), all lysozyme types exhibit similarity in their three-dimensional structure (Callewaert and Michiels, 2010), have nearly identical pH optima for their antimicrobial activity (Thammasiririak et al., 2007) and comprise the majority of hydrolytic enzymes in avian egg whites (Sun et al., 2017). We can therefore exclude potential bias resulting from using purified chicken lysozyme and ovotransferrin in our study.

The experiment was repeated 6 times with quail eggs and 5 times with pigeon eggs (i.e. in total, 360 quail eggs and 180 pigeon eggs were processed in 6 and 5 experimental sets, respectively; see Fig. S1 for details).

In each experimental set, the length and width of each experimental egg was measured by a digital calliper (Kinex, Prague, Czech Republic) with 0.01 mm accuracy to calculate egg volume. Then, eggshells were cleaned with 70% ethanol to completely remove the original eggshell microbiota, and the eggshell of each egg was gently perforated (controlling perforation depth) at approximately 5 mm below the equatorial region towards the blunt end with a 0.7×40 mm needle (Terumo, Eschborn, Germany) and injected using a 1 ml insulin syringe with 50 μl of AMP solutions (L+, O+, L+O+) dissolved in sterile PBS (GIBCO, pH 7.2, Invitrogen, Waltham, MA, USA). Each AMP solution corresponded to 20% of the species-specific mean protein concentration in eggs of our model species (Table 1). Control eggs were processed identically but injected with 50 μl of sterile PBS only. Needle perforations in the eggshells were sealed using a gel-based adhesive (Loctite-Super Attack, Henkel, Rocky Hill, CT, USA; Javurkova et al., 2015) and eggs were then placed into an incubator with automatic egg turning (OvaEasy 190 Advance, Brinsea Products Inc., Weston-super-Mare, UK) for incubation treatments. Whereas ‘full’ eggs were continually incubated for 4 days (96 h) at 37°C and 60% humidity, ‘partial’ eggs were incubated for 4 h day⁻¹ for 4 days (16 h in total) at 37°C and 60% humidity and kept for 20 h day⁻¹ for 4 days (80 h in total) at 10°C and 60% humidity in a cooling box, which corresponds to the mean daily temperature for both model species during the reproduction period under natural conditions. Finally, ‘no’ eggs were kept for 4 days (96 h in total) in a cooling box at 10°C and 60% humidity.

The experiment was conducted with the approval and under the supervision of the Ethical Committee of the Faculty of Science, Charles University, Prague (permit no. 13060/2014-MZE-17214).

After experimental procedures, each egg was cracked under sterile conditions in a laminar flow box and the egg white and yolk were carefully separated. Egg white pH was immediately measured using a pH meter with a Volcraft probe (PH-100 ATC), and egg whites were then transferred into 10 ml sterile plastic vials (Axygen™). Analyses of egg white antimicrobial activity and post-experimental AMP concentration were conducted immediately after the experiment. Egg fertility was assessed post-experimentally based on the presence of a cicatricule in the egg yolk using the method of Sellier et al. (2005). Post-experimental concentrations of egg white lysozyme and ovotransferrin were analysed using the methods described above.

The antimicrobial activity of egg white was measured by the agar well diffusion method (Arendrup et al., 2017; Voidarou et al., 2011) using the potentially pathogenic Micrococcus luteus (CCM 169; Pekala et al., 2018) and beneficial probiotic Bacillus subtilis (CCM168; Forte et al., 2016a,b; Li et al., 2016; Bai et al., 2017) obtained from the Czech National Collection of Type Cultures (CNCTC). Both indicator microorganisms have been detected in the egg content of naturally incubated eggs (Cook et al., 2003; Wang et al., 2011). Pilot experiments using six tested bacterial strains revealed that these indicator microorganisms are most sensitive to experimental precocial and altricial egg whites. For testing of antimicrobial activity, B. subtilis and M. luteus from agar slant cultures were dissolved in LB medium and incubated at 37°C under agitation of 180 rpm for 17 h. Then, cultures were dissolved to OD=5.0 and 100 μl of this suspension was added to 3 ml sterile top agar (0.7% LB agar) and poured onto 2% LB agar plates. Egg white samples (20 μl) were then placed in duplicate into 5 mm holes punched into the solidified agar plates. In addition, in each plate, one hole contained 20 μl of PBS and one hole contained lyophilized hen egg white lysozyme (62971, Sigma-Aldrich, at 10 mg ml⁻¹ concentration) as a negative and positive control, respectively. Plates were incubated for 24 h at 37°C. Then, images of plates with clearance zones around the holes were taken in a standard position with a Canon camera (EOS450D, 50 mm micro objective, F2.8). Areas of clearance zones were analysed using ImageJ 1.42q software (Schneider et al., 2012).

First, it is worth noting that in accordance with a previous study documenting strong differences in egg white antimicrobial properties in fertilized and un-fertilized eggs (Guyot et al., 2016), we found a significant effect of egg fertility on all tested variables [P<0.001 for ‘egg fertility’ in all general linear models (GLMs)]. We therefore used datasets including only fertile experimental eggs (n_precocial=325, n_altricial=61) for all subsequent statistical analyses.

Four separate GLMs with Gaussian error distribution were used to analyse the effects of incubation pattern (‘no’, ‘partial’, ‘full’), AMP treatment (control, L+, O+, L+O+), egg white pH and egg volume on the overall antimicrobial activity of altricial and precocial egg whites against M. luteus and B. subtilis. Incubation pattern, AMP treatment, egg white pH, egg volume and two-way interaction of AMP treatment and incubation pattern were included as explanatory variables in each GLM. Data for mean antimicrobial activity of altricial and precocial egg whites against B. subtilis were log-transformed to achieve normality.

To test factors responsible for changes in egg white pH, we used two separate GLMs including egg white pH of precocial and altricial eggs, respectively, as response variables, and AMP treatment, incubation pattern and their two-way interaction as explanatory variables.

Factors affecting post-experimental concentration of egg white AMPs were tested using four separate GLMs, where log-transformed post-experimental concentrations of egg white lysozyme and ovotransferrin measured separately for precocial and altricial eggs were included as response variables, and AMP treatment, coded as either manipulated (i.e. in the case of lysozyme, this category included L+ and L+O+ eggs and in the case of ovotransferrin it included O+ and L+O+ eggs) or control (i.e. no injection with AMPs), incubation pattern and their two-way interaction were set as explanatory variables.

Backwards-stepwise elimination of non-significant terms in the GLMs using the drop1 function were performed to select the best minimal adequate model (MAM; Crawley, 2007). Tukey's honest significant difference (Tukey's HSD) tests were used for multiple comparisons of significant effects and their means between tested categorical variables. All analyses and graphics were performed in the software RStudio (version 1.1.463; http://www.rstudio.com/) using the packages multcomp and ggplot2.

We found incubation patterns significantly yet selectively affected the egg white antimicrobial activity against M. luteus and B. subtilis in precocial and altricial eggs (Table 2). Against our expectations, partial incubation of precocial and altricial eggs did not alter their egg white antimicrobial activity against M. luteus compared with un-incubated eggs (Tukey's HSD: P=0.363 and P=0.890 for precocial and altricial eggs, respectively; Fig. 1A,B). Fully incubated precocial and altricial eggs significantly decreased their antimicrobial activity against M. luteus compared with un-incubated and partially incubated eggs (Fig. 1A,B). However, partial incubation reduced antimicrobial activity against B. subtilis in precocial eggs compared with un-incubated eggs (Tukey's HSD: P<0.001; Fig. 1C). Nevertheless, no incubation pattern had an effect on egg white antimicrobial potential against B. subtilis in altricial eggs (Table 2, Fig. 1D).

Concentrations of particular egg white AMPs had significant yet species-specific effects on antimicrobial potential against M. luteus and B. subtilis (Table 2). We found that precocial and altricial eggs with increased concentrations of egg white lysozyme had the highest antimicrobial activity against M. luteus compared with control eggs (Tukey's HSD: P<<0.0001 and P<0.01 for precocial and altricial eggs, respectively), followed by eggs with simultaneously increased concentrations of egg white lysozyme and ovotransferrin (Tukey's HSD: P<<0.0001 and P=0.011 for precocial and altricial eggs, respectively; Fig. 2A,B). Increased concentrations of egg white ovotransferrin did not affect antimicrobial activity of either precocial or altricial eggs against M. luteus (Tukey's HSD: P=0.724 and P=0.518 for precocial and altricial eggs, respectively; Fig. 2A,B). The antimicrobial potential of altricial and precocial eggs against the potentially beneficial B. subtilis was species specific. While increased egg white lysozyme in precocial eggs significantly inhibited the growth of B. subtilis compared with control eggs (Tukey's HSD: P<<0. 001; Fig. 2C), no AMP treatment had a significant effect on antimicrobial activity against B. subtilis in altricial eggs (Table 2, Fig. 2D).

Finally, we documented that egg white pH had a selective and species-specific effect on the antimicrobial activity of egg white against M. luteus and B. subtilis. While antimicrobial activity against M. luteus increased along with increased egg white pH in precocial eggs (Table 2, Fig. 3A), egg white pH did not underlie antimicrobial activity against M. luteus in altricial eggs (Table 2, Fig. 3B). In contrast, egg white pH had no effect on antimicrobial potential against B. subtilis in either precocial or altricial eggs (Table 2).

Consistent with the species-specific relationships between egg white pH and antimicrobial activity against M. luteus (Fig. 3A,B), we documented species-specific effects of incubation patterns on changes in egg white pH (Table 3). While the egg white pH of partially incubated precocial eggs significantly increased compared with that of un-incubated eggs (Tukey's HSD: P=0.002; Fig. 3C), the egg white pH of fully incubated precocial eggs significantly decreased compared with that of un-incubated and partially incubated eggs (Tukey's HSD: P<<0.0001 for un-incubated and partially incubated eggs; Fig. 3C). In contrast, incubation patterns had no significant effect on egg white pH in altricial eggs (Table 3, Fig. 3D). AMP treatment had no significant effect on egg white pH in either species (Table 3).

Incubation patterns significantly affected post-experimental concentrations of egg white AMPs (Table 3). In the case of lysozyme, this effect was species specific. We found that regardless of AMP treatment, egg white lysozyme concentration in precocial eggs significantly decreased in eggs undergoing full incubation compared with partially incubated and un-incubated eggs (Tukey's HSD: P<<0.0001 and P<<0.0001, respectively; Fig. 4A). In contrast, we found no effect of incubation pattern on concentrations of egg white lysozyme in altricial eggs (Table 3, Fig. 4B).

Unlike lysozyme, concentrations of egg white ovotransferrin increased under partial and full incubation compared with control un-incubated eggs in both precocial and altricial eggs (Tukey's HSD: P<0.0001 and P<0.01, respectively; Fig. 4C,D). However, we observed an interaction between incubation pattern and AMP treatment explaining the variability in post-experimental ovotransferrin concentration in precocial and altricial eggs (Table 3). A graphical representation of this interaction showed that incubation patterns affected egg white ovotransferrin concentration differently depending on initial ovotransferrin concentration in eggs. In particular, the effect of partial and full incubation increasing ovotransferrin concentration in precocial eggs was stronger in eggs with experimentally increased concentrations of ovotransferrin, whereas in altricial eggs, partial and full incubation increased ovotransferrin concentration only in control eggs (no addition of egg white ovotransferrin; Fig. 4C,D).

Finally, AMP treatment significantly increased post-experimental concentrations of egg white lysozyme in experimental eggs (Table 3). Lysozyme concentration in lysozyme-supplemented eggs (i.e. L+ and L+O+ eggs) was significantly increased by 2.19 mg ml⁻¹ in precocial eggs (Tukey's HSD: P<<0.001) and by 1.37 mg ml⁻¹ in altricial eggs (Tukey's HSD: P=0.002) compared with control eggs (Fig. 4A,B). Similarly, ovotransferrin concentration in altricial eggs increased by 1.03 mg ml⁻¹ in ovotransferrin-supplemented eggs (i.e. O+ and L+O+) compared with control eggs (Fig. 4D). However, precocial ovotransferrin-supplemented eggs had lower ovotransferrin concentrations, which decreased by 0.14 mg ml⁻¹ compared with control eggs (Tukey's HSD: P=0.043; Fig. 4C).

In this study, we experimentally demonstrated on eggs of two representative species from each end of the avian altricial–precocial spectrum, that different incubation patterns have selective and species-specific effects on antimicrobial protection of the developing embryo. Moreover, we found that these differences are most probably conditioned by temperature-mediated changes in egg white antimicrobial components that varied interspecifically along with incubation patterns.

Against our expectations, we found that full incubation of eggs decreased their antimicrobial capacity against a potentially pathogenic microorganism, whereas the antimicrobial activity against a beneficial probiotic bacterium remained unaffected in both precocial and altricial eggs. Moreover, partial incubation did not reduce the levels of pathogenic microorganism in the egg white of altricial and precocial eggs, but significantly decreased the antimicrobial effect of egg white against the beneficial probiotic microorganism in precocial eggs. These findings are in line with our previous study demonstrating no effect of partial incubation on the probability and intensity of microbial trans-shell infection in mallard (Anas platyrhynchos) eggs (Javurkova et al., 2014). It appears, therefore, that the protective role of incubation is inherent in the reduction of the eggshell microbial load (Cook et al., 2005a; D'Alba et al., 2010a; Ruiz-De-Castaneda et al., 2011, 2012; Soler et al., 2015; Bollinger et al., 2018) and the enhancement of commensal/beneficial eggshell microbes (Shawkey et al., 2009; Grizard et al., 2014, 2015) that are able to enter the egg interior. We demonstrated that the partial incubation of precocial eggs mitigated the ability of the egg white to reduce the levels of beneficial probiotic microorganism. This suggests that the protective role of partial incubation might be a complex process favouring the proliferation of beneficial/commensal microbes on the eggshell (Grizard et al., 2014, 2015) and then in the egg interior after their trans-shell penetration. It is worth noting, however, that these findings on the role of incubation patterns in affecting the antimicrobial activities of egg whites against the selected indicator microorganisms were most probably conditioned by strong species-specific effects of incubation patterns on egg white antimicrobial components.

We found that egg white pH of partially incubated precocial eggs significantly increased, whereas egg white pH of fully incubated precocial eggs decreased. Moreover, this decrease in egg white pH was related to antimicrobial activity against the potential pathogen in precocial eggs. This is in accordance with previous studies on precocial eggs documenting a short peak of egg white pH in freshly laid un-incubated eggs, which then gradually decreases toward neutrality during incubation as a result of the production of CO₂ by the developing embryo (Cunningham, 1974; Lapao et al., 1999; Fang et al., 2012a). Moreover, our results corroborate with studies that found strong antimicrobial action of alkaline egg white pH per se (Tranter and Board, 1984; Kang et al., 2006), or via the effect of alkaline pH enhancing antimicrobial activity of AMPs (Ibrahim et al., 1994, 2001; Halbrooks et al., 2005). However, we did not observe changes in egg white pH in altricial eggs resulting from different incubation patterns. As a similar trend with a rather slight decrease of egg white pH along with incubation cycle has been documented in studies on altricial eggs of the family Alaudidae (Horrocks et al., 2014; Grizard et al., 2015), this suggests that the effect of incubation pattern on egg white pH is likely to be species specific across the avian precocial–altricial spectrum.

Furthermore, our results demonstrated species-specific effects of incubation patterns on egg white lysozyme. We found that egg white lysozyme concentration decreased along with the intensity of incubation in precocial eggs, whereas no effect of incubation pattern was observed on lysozyme concentration in altricial eggs. At present, only a few studies have examined the relationship between egg white AMP concentration and incubation temperature, and in the case of lysozyme, their results are inconsistent. While studies investigating precocial chicken and altricial red-capped lark eggs reported a decrease in egg white lysozyme concentration along with incubation cycle (Cunningham, 1974; Grizard et al., 2015), Fang et al. (2012a,b) observed only a slight lysozyme decrease on the second day of incubation in chicken eggs. As the decrease in concentration of egg white lysozyme has been reported to be a result of binding to other proteins like ovomucin (Kato et al., 1975), the formation of protein complexes (Qiu et al., 2012; Liu et al., 2015) and its early degradation soon after the onset of incubation (Fang et al., 2012b), we suppose that the same processes were responsible for its decreased concentration in the egg whites of fully incubated precocial eggs in our study. Moreover, as pH may affect the physical properties of lysozyme and its ability to form complexes (Fang et al., 2012a; Qiu et al., 2012), we assume that the lack of variation in lysozyme concentration in altricial eggs as a result of incubation treatment might be explained by the lack of changes in egg white pH in incubated altricial eggs.

In contrast, we found that egg white ovotransferrin concentration increased in partially and fully incubated precocial and altricial eggs compared with un-incubated eggs. This is in line with previous studies (Fang et al., 2012a,b; Grizard et al., 2015) that have suggested an ovotransferrin increase as a result of water loss or disruption of the vitelline membrane during egg incubation. In addition, we found that the intensity of the ovotransferrin increase during incubation was dependent on its initial egg white concentration. While the largest increase in precocial eggs was observed in eggs supplemented with ovotransferrin, we found the opposite trend in altricial eggs, which had the highest increase of ovotransferrin in control eggs with natural concentrations of egg white ovotransferrin.

Finally, we found that experimentally increased concentrations of two dominant egg white AMPs – lysozyme and ovotransferrin – had species-specific and selective effects on the antimicrobial potential of avian precocial and altricial eggs against potentially pathogenic and beneficial microorganisms. While an increased concentration of egg white lysozyme enhanced the antimicrobial potential against a pathogenic invader in both precocial and altricial eggs, the effect of increased egg white lysozyme against a beneficial microorganism was apparent only in precocial, not altricial, eggs. In contrast, precocial and altricial eggs with experimentally increased egg white ovotransferrin did not change their antimicrobial potential against either of the tested microorganisms. These counteractive effects of egg white AMP concentration were most probably related to the fact that lysozyme is primarily bactericidal whereas ovotransferrin is bacteriostatic. Bactericidal lysozyme usually hydrolyses the peptidoglycan wall at the surface of Gram-positive bacteria (Hughey et al., 1989), disrupts the bacterial membrane (Masschalck and Michiels, 2003) or acts via the induction of autolysins (Laible and Germaine, 1985; Ibrahim et al., 2001). In contrast, ovotransferrin has broad-range potential against various microorganisms via iron chelation (Garibaldi, 1970; Aguilera et al., 2003). Ovotransferrin has also been documented to be bactericidal, but only against particular bacteria such as Bacillus cereus (Baron et al., 2014) or Escherichia coli (Ibrahim et al., 2000). This is in line with our study, where ovotransferrin-supplemented altricial eggs tended to increase their antimicrobial action against B. subtilis. It seems, therefore, that apart from the primary physiological roles of ovotransferrin during embryo development (Giansanti et al., 2012), its increased deposition may just reduce the proliferation of most invading microbes because of its limited bactericidal potential. In contrast, egg white lysozyme most likely has a primary antimicrobial role protecting the egg interior against pathogenic or potentially pathogenic Gram-positive bacteria. Increasing its physiological dose in precocial and even altricial eggs with naturally low lysozyme concentrations seems to have a strong protective function in un-incubated eggs or clutches with delayed incubation. Furthermore, the antimicrobial activity of egg white considerably increased in both precocial and altricial eggs supplemented simultaneously with lysozyme and ovotransferrin. This may either indicate that the two major egg white AMPs work in synergy or support the dominance of lysozyme, which overshadowed the effect of ovotransferrin. Regardless, as different microorganisms vary in their sensitivity to the antimicrobial action of particular egg white AMPs (Wellman-Labadie et al., 2008b; Bedrani et al., 2013b), it is highly probable that concentrations of particular AMPs reflect the adaptation of individuals or species to actual microbial pressure on their clutches (Grizard et al., 2015; Horrocks et al., 2015; Boonyarittichaikij et al., 2018). This idea has been supported experimentally (Bedrani et al., 2013a,b) and by studies that have demonstrated the increased deposition of lysozyme in eggs of species nesting in humid microbial-rich cavities (Wellman-Labadie et al., 2008c), suffering from more intensive eggshell microbial loads (Boonyarittichaikij et al., 2018), in the first egg of the clutch, which has a prolonged period of exposure to environmental microbes (Saino et al., 2002), and in eggs produced after copulation with more attractive males (D'Alba et al., 2010b).

It is therefore obvious that many other factors such as ambient temperature (Saino et al., 2004; Horrocks et al., 2014), nest humidity (Wellman-Labadie et al., 2008c), laying sequence and stage of breeding season (Saino et al., 2002; Bonisoli-Alquati et al., 2010) or individual intrinsic factors (D'Alba et al., 2010b) affect the deposition and/or activity of egg white AMPs. As we used two representative altricial and precocial model species bred under relatively controlled conditions, we can exclude these potentially confounding factors in our study. Yet, it is necessary to take into account possible co-variation of these factors affecting the deposition and/or activity of egg white AMPs with altricial–precocial developmental strategies in future studies focusing on the effect of incubation pattern on egg internal antimicrobial properties in free-living birds.

To conclude, our results indicate that changes in the concentration of egg white AMPs are probably an adaptation providing broad-range and/or selective antimicrobial protection of altricial and precocial eggs. Moreover, although we did not find a direct effect of incubation on improvement in the internal protection of altricial and precocial eggs of our model species against a pathogenic microorganism, it is highly probable that the antimicrobial protective role of avian incubation is broad range and linked with temperature-dependent changes in egg white pH and AMP concentration during incubation (Guyot et al., 2016). Although many factors such as ambient temperature or predation have been documented to fundamentally contribute to incubation pattern alterations (Matysioková and Remeš, 2018), our study highlights that variability in avian incubation patterns might be a behavioural mechanism specifically affecting antimicrobial properties of the egg interior.

Many thanks to Dr Jakub Kreisinger for statistics and experimental design consulting, to Dr Jana Beranová and Lucie Jánská for their help with microbiological analyses and laboratory assistance, to Oldřich Mach, Petr Smrček and Stanislav Česnek for their help with quail and pigeon breeding and experimental egg collection, to Dr Michal Vinkler and Dr Zuzana Świederská for sharing the space in experimental facilities and to Dr David Hardekopf for English proofreading and editing the manuscript.