Parasite-host coevolution can lead to novel behavioural adaptations in hosts to resist parasitism. In avian obligate ­­­brood parasite and host systems, many hosts species have evolved diverse cognitive and behavioural traits to recognize and reject parasitic eggs. Our understanding of the evolution and ecology of these defences hinges on our understanding of the mechanisms that regulate them. We hypothesized that corticosterone, a hormone linked to the stress-response, vigilance, and the suppression of parental behaviour, stimulates the rejection of foreign eggs by brood parasite hosts. We experimentally reduced circulating glucocorticoid levels with mitotane injections in American robins Turdus migratorius and found that the mitotane-treated birds rejected foreign eggs at a lower frequency compared to the sham-treated subjects. This is the first study to causally identify a potential mechanism of a widespread defence behaviour, and it is consistent with egg rejection being mediated by stress physiology.

Methods

1. Field site and species

We studied wild American robins Turdus migratorius, an occasional host to obligate brood-parasitic brown-headed cowbirds Molothrus ater, in Urbana, IL, USA, during the summer of 2019 (see details of the study area in [15,16]). For this study, we focused only on female robins, because it is the sex responsible for egg rejection in this species [17]. 

2. Treatment validation

Hormonal implants that result in supraphysiological hormone levels may result in ecologically irrelevant phenotypes [18]. We therefore opted to suppress glucocorticoid levels using mitotane, a glucocorticoid synthesis inhibitor, which has been shown to consistently reduce both baseline and stress-induced [19–22] corticosterone levels in birds. We first tested if the directional effect of mitotane on corticosterone in American robins parallels that already seen in other songbird species. We caught wild egg-laying or incubating robin (n=8) females and took a baseline blood sample from the brachial vein within 3 min of capture (mean start time = 130 sec; mean end time = 167 sec). Blood was stored on ice and centrifuged at 8000 RPM within 2 hrs to separate plasma. Plasma samples were kept on ice until frozen at –80 °C 4 hrs later. We then injected the pectoral muscle of 4 females with 34 mg mitotane (Sigma-Aldrich, Cat. No. 25925), dissolved in 400 µl sterile peanut oil (Acros Organics, Cat. No. 416855000, dosage 400 mg/kg), following guidelines for a high mitotane dosage in songbirds [19,20]. Four sham females were injected with peanut oil vehicle (400 µl). 

In our population, robins become wary of mist nets and humans after capture. We therefore temporarily moved these birds into captivity, housing them singly overnight in 40×40×34 cm cages, and providing ad libitum water, earthworms, bananas, and crushed dog food. The following day, we again collected their blood within 3 min of capture (mean start time = 94 sec; mean end time = 120 sec). 

Repeated administrations of mitotane can have adverse effects, such as lethargy, in the long term [23,24], but this effect has not been observed in birds using single injections [19-22], especially regarding self-maintenance behaviours within a 24 hr period [22]. Nonetheless, to assess the possibility of unintended side effects, we observed if mitotane treatment affected feeding behaviour in captivity by measuring the birds’ mass (nearest g; see Section (d) below for field-based assessment of possible lethargy). 

To test the effect of mitotane on glucocorticoid levels, we analysed plasma corticosterone using an enzyme immunoassay (Cayman Chemical, Cat. No. 501320). Validation details and methods for this assay using robin plasma are published elsewhere [25]. All samples were assayed in duplicate on the same plate (intraplate coefficient of variation = 6.7%). 

3. Hormone manipulation of wild birds

We captured incubating robin females (n=65) at their nests using a mist net between 6-10 am, after they had completed their clutches (median 2 days, range 0-5 days after clutch completion [26]). We first took a 450 µl blood sample as part of a different study. We then treated each bird randomly either with mitotane (n=38) or sham (n=27), as described above. We also took standard morphometric measurements, including age [27], mass (nearest g), and tarsus (nearest 0.1 mm). We fitted birds with a USGS band and 3 colour bands (Avinet), following which the birds were released. An unanticipatedly large number of birds abandoned their nests after the treatments (see Results and the Ethics statement).

4. Experimental parasitism

American robins reject the majority of natural cowbird [28] or cowbird-like model eggs [3], but they show variable responses to egg colours near their rejection threshold [29]. Importantly, robins also show intermediate and individually repeatable rejection rates to deep-blue cowbird-sized model eggs (figure 2 inset; for details see [30]). We therefore used deep-blue, 3D-printed eggs to investigate the effect of the injection treatment on egg rejection.

A day after the mitotane or sham injections, we added one deep-blue model egg to the nest. We did not remove any robin eggs, because Turdus thrushes show the same response to model eggs regardless of whether their own eggs are removed [31]. We returned to the nest one day later to record whether the model egg was accepted (present) or rejected (missing). On each visit, we verified the identity of the female using band colours. If the female was absent during these visits, we returned to the nest later to confirm female identity or nest abandonment (cold eggs). One female died and two nests were depredated and these were excluded from further analyses.

To assess if mitotane caused general loss of motor activity in the wild, we quantified vigilance behaviour and nest abandonment. We recorded vigilance before and after the injection by approaching the nest at a steady pace, typically from 20 m or greater distance, and recording the distance at which birds took flight (flight initiation distance (FID), measured by counting paces [32,33]). We also noted whether mitotane- or sham-treated birds differed in the probability of nest abandonment.

5. Statistical analyses

Hormone and mass data in the validation study, and FID data from the field experiment, were not normally distributed, therefore we used Mann-Whitney U-tests to assess the differences in these variables between treatments. 

Life history, seasonal, and morphological variables did not differ between the treatment groups (all p>0.05, supplementary table 1). Because treatments were randomized across individuals and time, we assessed the effect of experimental injections on categorical behaviours (yes/no nest abandonment and egg rejection) using Fisher’s exact tests (a=0.05). Clutch size, previously associated with individual variation in egg rejection [25], did not explain significant variation in egg rejection in these data (supplementary table 2), and is not included in analyses.