Introduction

Obligate brood parasitism refers to a specific reproductive behavior exhibited by certain animals, such as ants, fish, amphibians, and birds, where they lay their eggs in the nests of other species, and the host provides parental care for the offspring of the parasite (Davies 2000; Kilner and Langmore 2011). Parasitic birds, such as cuckoos, impose strong selective pressures on their hosts, leading hosts to evolve a range of anti-parasitic strategies. These strategies include attempting to prevent cuckoos from approaching the nest, resisting parasitism by causing a commotion, attacking cuckoos, rejecting or burying foreign eggs in the nest, or removing parasitic nestlings, or deserting the nest (Kilner and Langmore 2011; Feeney et al. 2012; Guigueno et al. 2014; Soler 2014). These defensive strategies employed by hosts, in turn, drive parasites to evolve more effective behaviors to enhance their chances of successfully deceiving the hosts (Davies 2011; Soler 2017; Rojas Ripari et al. 2021). This creates a co-evolutionary arms race between parasites and hosts (Feeney et al. 2014). Currently, research on the co-evolutionary arms race between parasites and hosts primarily focuses on the mimicry of parasite egg and nestling morphology, and the increasingly complex cognitive and behavioral responses of hosts to parasite eggs and nestlings (Davies 2011; Soler 2017; Rojas Ripari et al. 2021). Exploring the complex egg rejection behavior in host birds provides an excellent opportunity to understand animal cognitive abilities, as egg rejection necessarily involves cognitive mechanisms related to the recognition of parasitic eggs (Rothstein 1974; Hanley et al. 2021).

Among the various mechanisms that host birds use to recognize and reject foreign eggs, two hypotheses have been widely discussed (Feeney et al. 2014; Manna et al. 2017). The first is template-based recognition, also known as true recognition, where hosts reject foreign eggs based on a genetic or learned memory template (Rothstein 1975; Hauber and Sherman 2001; Wang et al. 2015; Yang et al. 2022). Hosts possessing this recognition mechanism can reject foreign eggs even when there are no eggs of their own in the nest for comparison (Bán et al. 2013; Yang et al. 2014; Yi et al. 2020). Another mechanism is recognition by discordancy, where hosts directly compare their own eggs in the nest with the – potentially – foreign eggs to reject the most dissimilar ones (Rothstein 1975; Servedio and Lande 2003; Yang et al. 2014). Hosts using the mechanism of recognition by discordancy typically reject a minority of eggs that differ from most eggs in the nest, even if the minority of eggs are their own eggs (Rothstein 1974; Moskát et al. 2008, 2014b; Yang et al. 2014).

In addition to the two egg recognition mechanisms mentioned above, some studies suggest that host birds may employ other types of egg recognition mechanisms. For example, the “onset of laying mechanism” is used by hosts who consider any egg appearing in the nest before they lay their own eggs as foreign and will reject them (Sealy 1995; Davies 2000). The “phenotype distribution mechanism” is a relatively unexplored variant of template-based recognition, specifically focused on clutch assessment through template updating. The “phenotype distribution mechanism” posits that hosts observe the appearance of their own eggs to determine the distribution characteristics of their eggs. They reject eggs that fall outside the range of their own egg phenotype distribution (Servedio and Lande 2003; Hanley et al. 2017), following a decision rule shaped by flexible optimal acceptance thresholds (Holen and Johnstone 2006; Moskát and Hauber 2007). Hosts are more likely to exhibit greater phenotypic variation in their own eggs as the number of eggs laid increases(Reeve 1989). This potentially causes the characteristics of parasitic eggs to fall within the expanded range of host egg characteristics. Consequently, a host that initially rejected parasitic eggs may shift to accepting them (Hauber et al. 2006; Holen and Johnstone 2006; Moskát and Hauber 2007). This means that hosts employing the phenotype distribution mechanism may accept parasite eggs falling within their own egg phenotype distribution range, even if the parasite eggs are in the minority.

Once parasitic eggs are recognized, hosts may employ various strategies to reject them, including ejecting, deserting, or burying the eggs (Davies and Brooke 1989; Guigueno et al. 2014; Sealy and Underwood 2012; Ma and Liang 2021). However, owing to variations in host age, breeding stage, perceived parasitic pressure, co-evolutionary history with parasites, and the costs associated with rejection, hosts may exhibit plasticity in their response to parasitic eggs (Davies et al. 1996; Thorogood and Davies 2013; Moskát et al. 2014a; Soler 2017; Martínez et al. 2020; Liu et al. 2021; Zhang et al. 2021, 2022). Some studies have indicated that host rejection rates of parasitic eggs can change during different stages of the breeding cycle (Moskát and Hauber 2007; Soler et al. 2015; Wang et al. 2015; Ma et al. 2024). However, other studies have shown that host egg rejection behavior can remain stable and repeatable (Samaš et al. 2011; Grim et al. 2014). Therefore, the impact of different breeding stages on host egg rejection is inconsistent across studies.

Liu et al. (2023) conducted a study on azure-winged magpies (Cyanopica cyanus) and concluded that these birds employ a true recognition mechanism for egg recognition. However, in their experiments, azure-winged magpie eggs were either in the majority in the nest or equal in number to the experimental quail eggs (Coturnix japonica). They lacked an experimental group where azure-winged magpie eggs were fewer than the quail eggs, making it impossible to completely rule out the mechanism of recognition by discordancy. To elucidate the egg recognition mechanism of azure-winged magpies and determine if their egg rejection behavior varies across different breeding stages (pre-egg-laying, one-host-egg and multi-host-egg stages, and early incubation period), we conducted a field experiment on two azure-winged magpie populations in Fusong, Jilin and Wuhan, Hubei, China. These two populations are separately parasitized by different species and experience different parasitic risks. In Fusong, azure-winged magpies are most likely to be parasitized by Indian cuckoos (Cuculus micropterus), because in Zhangqizhai village (41°30′ N, 123°45′ E), Benxi, Liaoning, located 300 km away from our study site in Fusong, azure-winged magpies were found to be parasitized by Indian cuckoos. However, we did not find any azure-winged magpie nests being parasitized (n = 112) in Fusong. In Wuhan, azure-winged magpies were parasitized by Asian koels (Eudynamyss scolopaceus) (parasitism rate: 5.4%, n = 258), and most nests were multiple parasitized by Asian koels. Asian koel eggs are slightly larger than azure-winged magpie eggs, but to the human eye, they are highly mimetic to azure-winged magpie eggs in both egg coloration and spotting. We hypothesized that the rejection rate of parasitic eggs in Wuhan will be higher than that in Fusong. It is believed that the strength of anti-parasite defense is positively correlated with the risk of parasitism (e.g. Stokke et al. 2008, but see Samas et al. 2014), which is particularly high during a host’s egg laying stage but decreases or absent after early incubation (Moskát and Honza 2002; Moskát 2005). Therefore, we hypothesized that if hosts use the onset of laying mechanism to reject all eggs that are laid in nests prior to the onset of their own laying, we predicted high rejection rates in the pre-egg laying stage, and lower rejection rates during the remainder of laying stages. And if hosts used the phenotype distribution mechanisms for egg recognition, we predicted a low rejection rate of parasite eggs during multi-host-egg stages and during the pre-laying stage. Finally, if hosts used the mechanism of recognition by discordancy, we predicted a higher rejection rate of parasite eggs during late than early egg laying. In contrast, if hosts only used the true recognition mechanism, we predicted high rates of rejection irrespective of the stage of the laying cycle.

Materials and methods

Study area and study species

Fusong is located in the southeastern part of Jilin, China (42°19ʹ N, 127°15ʹ E), and falls within a cold temperate and humid climate zone. This region experiences four distinct seasons, with long, cold winters and rainy summers (Liu et al. 2023). In Fusong, azure-winged magpies begin building nests in mid-April and lay eggs in late April and early May, with an average clutch size of 5.2 ± 2.4 eggs (range: 2–11, n = 112).

Huangpi (30°42ʹ N, 114°10ʹ E) is a district of Wuhan, China, located on the northern bank of the middle reaches of the Yangtze River. In Wuhan, most azure-winged magpies begin building nests in early May and lay eggs in mid-May, with an average clutch size of 5.4 ± 1.5 eggs (n = 31). However, some individuals also start laying eggs at the end of June and early July. Since we have not put colored rings on the azure-winged magpies, we cannot know whether there is a phenomenon of secondary reproduction in the azure-winged magpies in Wuhan. In Wuhan, the phenomenon of Asian koel eggs being rejected by azure-winged magpies was not observed in our study. Additionally, the nests of Azure-winged magpies, which were parasitized by Asian koels, were excluded from the egg recognition experiments.

Field experiment

From April to July 2023, we conducted a systematic search for the breeding nests of azure-winged magpies in the study area to perform egg recognition experiments on their nests.

Egg recognition experiments at different nesting stages

In Fusong, we categorized the nests of azure-winged magpies we found into three groups: (1) the pre-egg-laying group: azure-winged magpies had completed nest construction but had not yet started laying eggs; (2) the one-host-egg group: azure-winged magpies had just laid one egg in their nests; (3) the multi-host-egg group: azure-winged magpies had already laid multiple eggs in their nests (usually 2–3 eggs, less than average clutch size). For these nests, we added one red model egg directly to the nest (Fig. 1) (Table 1). In Wuhan, since the height of the nests of azure-winged magpies is generally more than 10 m, it is difficult to find enough nests to conduct experiments, so we only conducted egg recognition experiments in the pre-egg-laying stage. For the pre-egg-laying group, we directly added one blue model egg to the nest. For all experimental nests, we observed the responses of azure-winged magpies to the experimental eggs for 6 days (Moksnes et al. 1991). If the azure-winged magpies did not reject the egg and there were no peck marks on it, it was recorded as accepted. If the egg disappeared or had peck marks on it, it was recorded as rejected (Liu et al. 2023). Nests that were preyed upon or damaged within 6 days were not included in the analyses. During the pre-egg-laying period, when we added the model egg, we mounted a mini-camera (Uniscom-T71, 70 mm × 26 mm × 12 mm; Mymahdi Technology Co., Ltd., Shenzhen, China) in an inconspicuous location above the nest to ensure that the disappearance of the eggs was not attributed to nest predation.

Fig. 1
figure 1

Egg recognition experiments in nests of the azure-winged magpie. (A) Addition of one red model egg to the nest during the pre-egg-laying stage; (B) Addition of one red model egg to the nest during one host egg stage; (C) Addition of one red model egg to the nest during multi-host-egg stages; (D) Experimental nests with four cockatiels eggs and four host eggs; (E) Addition of four cockatiels eggs to the nest; (F) Addition of one blue model egg to the nest during the pre-egg-laying stage; (G) Experimental nests with three cockatiels eggs and three host eggs; (H) Experimental nests with four cockatiels eggs and two host eggs

Full size image
Table 1 Summary of the frequencies of egg rejection by the azure-winged magpie host in this study
Full size table

Egg recognition mechanism experiments

In Fusong, after confirming that the azure-winged magpies recognized and discarded the blue model eggs, the egg recognition mechanism was further tested. In order to prevent azure-winged magpies from preserving the memory of the same blue model eggs, we used another type of eggs (i.e. commercially available pure white eggs of the cockatiel, Nymphicus hollandicus) different from the blue model eggs and azure-winged magpie eggs, to test the egg recognition mechanism of azure-winged magpies.We conducted the following two experiments. Experiment 1, the “4 + 4” experiment, when the azure-winged magpies stopped laying eggs, we controlled the number of azure-winged magpie eggs in the nest to be the same as the number of cockatiel eggs, which was four eggs each (Table 1). In this experiment, the average clutch size of azure-winged magpies was 5.1 ± 0.9 (range: 4–6, n = 15). Therefore, some real eggs of the azure-winged magpie were removed from the nests and placed in an artificial incubator. Since all the azure-winged magpies in this experiment rejected the cockatiel eggs within two days, the real eggs of the azure-winged magpie were returned to the original nests the following day and continued to be incubated by the same parent birds. The final hatching rate was unaffected by the experiment. Experiment 2, we directly added four cockatiel eggs to the azure-winged magpie nests all at once (Fig. 1). In this experiment, all azure-winged magpie nests had a clutch size exceeding four eggs (mean ± SD: 7.8 ± 1.6, range: 6–11, n = 18). Experiments 1 and experiments 2 were conducted on different nests.

In Wuhan, the number of azure-winged magpie nests was limited. Thus, there was not enough nests to conduct egg recognition experiments at different breeding stages. So we directly conducted experiments on egg recognition mechanisms for some nests. We also conducted two experiments when the nests had a full clutch of eggs. Experiment 1, the “3 + 3” experiment, where we controlled the number of azure-winged magpie eggs in the nest to be the same as the number of cockatiel eggs, which was three eggs each(Table 1). Experiment 2, the “2 + 4” experiment, where we controlled the number of azure-winged magpie eggs in the nest to be two, and the number of cockatiel eggs was four (Fig. 1). Experiments 1 and experiments 2 were also conducted on different nests.

The experimental model eggs used in egg recognition experiments at different nesting stages were made of blue or red soft clay for modeling (mean ± SD: 21.67 ± 0.95 × 16.32 ± 0.62 mm, 2.91 ± 0.29 g, n = 115), whereas the pure white cockatiel eggs used in egg recognition mechanism experiments were non-fertilized eggs purchased online (18.13 ± 1.15 × 15.10 ± 0.46 mm, 1.73 ± 0.40 g, n = 10). All types of experimental eggs were smaller in size than the azure-winged magpie eggs (26.08 ± 0.69 × 19.53 ± 0.41 mm, 5.15 ± 0.23 g, n = 20). In most studies on host egg recognition, the phenotype of experimental eggs differs significantly from that of host eggs. Therefore, the experimental eggs used in this study were either bright blue, bright red, or white. The phenotypes of these eggs differed from those of azure-winged magpie eggs. Several studies have demonstrated that the coloration of model eggs influences the host’s decision to reject them (Abolins-Abols al. 2019; Manna et al. 2020; Hanley et al. 2021), for instance, green-backed tits (Parus monticolus) exhibit a higher rejection rate towards blue model eggs compared to red ones (Liu et al. 2024). We planned to use two color models for egg recognition at both sites, but we did not find enough nests. So different colored model eggs were used at two study sites to test egg rejection rates in azure-winged magpies. Previous study have shown that azure-winged magpies incur a rejection cost when rejecting larger experimental eggs (Liu et al. 2023). Therefore, the size of the experimental eggs used in this study was generally smaller than that of the azure-winged magpie.

Statistical analyses

Statistical analysis was conducted using IBM SPSS version 20.0 for Windows. Fisher’s exact test or Pearson’s chi-squared test was used to compare the rejection probabilities between different experimental groups. All tests were two-tailed, and the statistical significance was set at P < 0.05.

Ethics and permits

The experiments comply with the current laws of China, where they were performed. Experimental procedures were in agreement with the Animal Research Ethics Committee of Hainan Provincial Education Centre for Ecology and Environment, Hainan Normal University (No. HNECEE-2014-005).

Results

In Fusong, there was a significant difference in the rejection rate of red model eggs among different nesting stages of the azure-winged magpie (Fisher’s exact test, χ2 = 6.90, df = 2, P = 0.017) (Table 1). During the one-host-egg stage, all egg rejection behaviors occurred before reaching the full clutch size. Different nesting stages of azure-winged magpies showed significant differences in their egg rejection methods. During the pre-egg-laying stage, 25% of rejection behaviors were achieved through nest abandonment, and the remaining instances involved egg ejection, whereas during the one-host-egg stage and the multi-host-egg stage, all egg rejection behaviors were carried out through egg ejection (χ2 = 6.289, df = 2, P = 0.039). During the incubation period, whether in the “4 + 4” experiment (n = 15), where the number of azure-winged magpie eggs in the nest equaled the number of cockatiel eggs (four eggs each), or in the experiment where four cockatiel eggs were directly added to the azure-winged magpie nest (n = 18), azure-winged magpies rejected cockatiel eggs 100% of the time without making errors, such as mistakenly discarding their eggs or pecking their eggs during the process of ejecting foreign eggs.

In Wuhan, azure-winged magpies exhibited a 100% rejection rate of the blue model eggs during the pre-egg-laying stage (n = 17), and all rejection behaviors occurred before azure-winged magpies laid their eggs. Among these, 76.5% of individuals abandoned the nest after rejecting the model eggs, and 11.8% of individuals without rejecting the model eggs eventually abandoned the nest. During the incubation period, whether in the “3 + 3” experiment (n = 16) or the “2 + 4” experiment (n = 15), azure-winged magpies rejected cockatiel eggs 100% of the time. All rejection behaviors involved egg ejection, and there were no rejection costs incurred during the rejection process.

Discussion

Our study demonstrates that azure-winged magpies exhibit varying rejection rates for non-mimetic eggs at different breeding stages. During the pre-egg-laying and multi-host-egg stages, azure-winged magpies almost always rejected parasitic eggs. However, during the one-host-egg stage, the rejection rate for parasitic eggs significantly decreaseds, indicating that the breeding stage influenced the egg rejection behavior of azure-winged magpies. Although some experiments were observed in only one of the populations, our study demonstrates irrespective of whether the number of azure-winged magpie eggs in the nest compared to the number of foreign eggs, azure-winged magpies in the incubation stage accurately rejected the experimental eggs. Therefore, our study suggests that azure-winged magpies are capable of true egg recognition, and their egg recognition mechanism is template-based.

During the pre-egg-laying stage, azure-winged magpies rejected almost all foreign eggs that appeared in the nest. At this stage, azure-winged magpies might utilize an innate or long-term memory-based template for true egg recognition (Moskát and Hauber 2007) or “the onset of laying mechanism” to identify foreign eggs. These two egg recognition mechanisms are not mutually exclusive. If azure-winged magpies use template-based recognition, even in the absence of their own eggs, they can still accurately reject parasitic eggs (Moskát et al. 2010n et al. 2013; Yang et al. 2014; Yi et al. 2020). If they employ “the onset of laying mechanism,” they are not restricted by whether the host can recognize foreign eggs. Even if the host cannot recognize parasitic eggs, they can still reject foreign eggs that appear in their nest during the pre-egg-laying experimental parasitism or natural nest parasitism situations (Stouffer et al. 1987; McRae 1995; Guigueno and Sealy 2009; Soler et al. 2015). Furthermore, during the pre-egg-laying stage, azure-winged magpies may also adopt simpler rules to reject foreign eggs that appear in the nest, considering anything present in the nest before they lay their eggs as foreign objects that should be removed (i.e., nest sanitation; Rothstein 1975; Moskát et al. 2003; Yang et al. 2015; Guigueno and Sealy 2017; Feng et al. 2019). Since our study did not include highly mimetic egg parasitism experiments as in Wang et al. (2015), we cannot definitively determine which mechanism azure-winged magpies used to recognize parasitic eggs during the pre-egg-laying stage.

The lower rejection rate of eggs during the one-host-egg stage compared to the pre-egg-laying and multi-host-egg stages is similar to the results of yellow-bellied prinias (Prinia flaviventris) (Wang et al. 2015). Azure-winged magpies, like yellow-bellied prinias, may employ a long-term memory-based template to recognize their eggs. However, in each breeding cycle, they may still update or reinforce their egg recognition template by inspecting and learning the phenotype of their own eggs (Lotem et al. 1995; Moskát and Hauber 2007; Strausberger and Rothstein 2009; Soler et al. 2013). Before they lay their eggs, azure-winged magpies lack the opportunity to learn the phenotype of their own eggs, as shown in the magpie (Pica pica) that females tested in their first breeding attempt always accepted the model egg, even those individuals whose mothers were egg rejecters (Molina-Morales et al. 2014). However, once they have laid their first egg, they can imprint on the phenotype of their eggs by inspecting them. Parasitizing after the host has laid its first egg may confuse or cause hesitation in some young individuals or those without breeding experience, leading them to accept non-mimetic eggs that are less mimetic (Wang et al. 2015). Parasitism during the multi-host-egg stages, even in young and inexperienced azure-winged magpies, allows them to imprint their eggs and form a memory-based template. Parasitism at this stage does not confuse azure-winged magpies between their eggs and parasitic eggs, resulting in a higher rejection rate of non-mimetic eggs compared to the one-host-egg stage (Rothstein 1975; Hauber and Sherman 2001).

Our results differ from the findings of Moskát and Hauber (2007), as their study on great reed warblers (Acrocephalus arundinaceus) showed a significantly higher rejection rate of foreign eggs during the one-host-egg stage compared to the multi-host-egg stages. This difference may be due to the different ability of the two hosts to recognize foreign eggs. The great reed warbler has a strong ability to recognize and reject highly mimetic eggs of conspecific species, whereas the azure-winged magpies we studied could not recognize conspecific eggs (Liu et al. 2023), so they used different egg recognition mechanisms. Great reed warblers use “the phenotype distribution mechanism,” relying on the phenotypic characteristics of their eggs to recognize and reject foreign eggs. In contrast, the azure-winged magpies in our study employ a template-based recognition mechanism.

During the incubation stage, both in Wuhan and Fusong, azure-winged magpies accurately rejected cockatiel eggs, regardless of whether the number of azure-winged magpie eggs in the nest was more than, less than, or equal to the number of cockatiel eggs. This indicates that azure-winged magpies employ a true egg-recognition mechanism, consistent with the conclusions of Liu et al. (2023). However, in this study, we did not find any rejection cost associated with azure-winged magpies rejecting foreign eggs, similar to the results of Son et al. (2015) but different from the results of Liu et al. (2023) and Avilés (2004). This difference may be due to the size of the model and cockatiel eggs used in our study, which were both smaller than azure-winged magpie eggs. This may have facilitated azure-winged magpies in rejecting foreign eggs through ejection. In contrast, Liu et al. (2023) and Avilés (2004) used quail eggs, which are not only larger but also have relatively thicker shells, making it less favorable for azure-winged magpies to pierce and reject them (Swynnerton 1918; Krüger 2011). Therefore, this may have contributed to differences in the observed rejection costs in azure-winged magpies across different studies.

In summary, our study indicates that although the egg recognition mechanism in azure-winged magpie populations in China is true recognition, different breeding stages, particularly the nesting stage and egg-laying period, can still influence the recognition and rejection responses of azure-winged magpies towards parasitic eggs. Since we did not conduct highly mimetic egg (i.e. conspecific egg) simulation experiments on azure-winged magpies at different nesting stages, further research is needed to explore the impact of mimetic eggs on the egg rejection responses and mechanisms at different breeding stages. In addition, the age of azure-winged magpies studied was not known, we still did not know whether learning occurs during the female’s first breeding attempt, or whether such learning can continue over successive clutches (Grim et al. 2014 and a review in that study), which needs to be confirmed in future work.