Harlequin ladybird (Harmonia axyridis) shows high level of colour polymorphism. Particular forms differ in their colour combination, pattern and abundance.

We tested the willingness of native, wild-caught passerines to attack and eat particular forms. We predicted that those forms that are more abundant in the wild are better protected, as they are more familiar to the predators. We compared the forms novemdecimsignata and spectabilis, which represent 97 % of individuals in the wild population, but importantly differ in their visual appearance (mostly orange vs. mostly black). Further, we examined three intermediate forms: axyridis, which is very scarce in the wild, and two forms not occurring in the wild – originating from laboratory breedings: suturalis and aninkae. As predators, we used great tit (Parus major), which is very aversive towards ladybirds, and tree sparrow (Passer montanus), which is quite willing to attack and even eat ladybirds. We compared their responses to particular ladybird forms, with artificially brown-painted ladybird to test for the effect of visual signal.

We showed that both bird species attacked all forms of ladybird equally and usually very scarcely. The brown-painted novemdecimsignata form was attacked more often, showing that the visual appearance prevents birds from attacking any conspicuous colour combination. Sparrows tended to eat the attacked ladybirds of forms novemdecimsignata, spectabilis and brown painted. Scarce axyridis form and laboratory forms were very well protected from the attack, very likely due to neophobia.

Prey

Firstly, we used form novemdecimsignata (otherwise often called succinea), which has orange-red elytra with 19 black spots on both (Fig. 1). This form is the most common in the wild populations of Czech Republic (88 % of population – Nedvědová et al. 2013). Second form was spectabilis with black elytra with two red spots on each (Fig. 1). This form is the second most common in Czech Republic (9 % of population - Nedvědová et al. 2013). The third form, axyridis is the most common in some countries of origin of this species (Belyakova 2012); nevertheless, in the Czech Republic it is very scarce (1:1000 individuals, Nedvěd 2014). It has black elytra with six red spots on each (Fig. 1). Besides, two forms not occurring in the wild were used. Suturalis form evolved through a spontaneous mutation in the breeding of equicolor form (also very scarce in the wild) in the laboratory breeding at the Faculty of Sciences, University of South Bohemia (FSci USB). It has red-and-black elytra with similar area of both colours, red being in the frontal part of elytra (Fig. 1). Aninkae form also evolved through a spontaneous mutation in the laboratory, in this case the maternal form was axyridis. It has a complex red-and-black pattern (Fig. 1).

Harlequin ladybird possess defensive chemical in their haemolymph. Two alkaloids (harmonine and 3-hydroxypiperidin-2-on) are responsible for toxicity of the ladybird body and volatile methoxypyrazines mediate the unpleasant chemical signal (Alam et al. 2002). Recent GC MS measurements of harmonine content (Nedvěd, unpublished) showed no differences between colour morphs but its high increase during ageing of ladybirds.

To test the effect of the colour signal per se, we altered the ladybird coloration using paint with brown water paint (Koh-i-noor Hardtmuth—natural sienna, Fig. 1). This paint is nontoxic and has been shown to have no impact on bird behaviour and act as a non-warning colour to passerines (Dolenská et al. 2009; Průchová et al. 2014; Veselý et al. 2017). As all ladybirds were presented alive during the experiment, we left several days after the paint to completely restore the level of toxin in their haemolymph (Holloway et al. 1991). At the same time, the odour of methoxypyrazines in the haemolymph adhered to the ladybird surface vanished during the first 24 h (Holloway et al. 1991).

All forms are kept as homozygous strains in a breeding facility at FSci USB. The funders of those strains occurring naturally in the wild of Czech Republic were collected in years 2008-2010 in the surrounding of České Budějovice. The ladybirds were reared at 25°C, 16L:8D photoperiod and then as adults were kept at 5⁰C and 12L:12D. They were provided with moist cotton wool and half a grapevine berry as a water and energy supply. The age of the adults was at least one month.

Predators

Birds of both species were adults captured in the wild at the winter feeding spot. Captures were conducted from September to April, outside the breeding season, in years 2009-2022. Each bird was tagged with an aluminium ring with unique code and transported in the captivity. Birds were kept in commercially sold birdcages with water, mealworms (larvae of Tenebrio molitor) and sunflower seeds available ad libitum under the natural daylight regime and 15 ⁰C. The birds were kept in captivity for two days prior to the experiment, to get accustomed to laboratory conditions, and released immediately after the experiment at the site of capture.

Experiment

The experiment was conducted in another cage made of wooden cubic frame (70 cm× 70 cm × 70 cm) with the front wall made of a one-way mirror enabling the observer to see detailed behaviour without disturbing the focal bird. The cage was equipped with a perch, a bowl of water, and a rotating circular feeding tray containing small white cups, where only the cup nearest to the perch contained one prey item during each trial. Standard illumination was obtained by a light source (LUMILUX COMBI 18W, OSRAM) that simulates full daylight spectrum including UV radiation.

Each bird was presented with only one ladybird form. The bird was placed into the cage several hours prior to the experiment to accustom it to the environment and was offered several mealworms to learn to search for them in the cups and to attack them. After being trained to attack the offered mealworm immediately, the bird was deprived of food for 2 h to emphasize its motivation to forage. The experiment consisted of a sequence of 10 successful consecutive 5-min trials, in which the bird was offered alternately a mealworm or a ladybird (sequence mealworm, ladybird, mealworm, ladybird, etc.). All five ladybirds presented to one bird was of the same form. The mealworm trials were performed to check for the motivation of birds to attack insect prey and were not further analysed. In all mealworm trials, the mealworm was attacked and eaten quickly after being offered.

We recorded six behaviours of the focal bird. Attacking was scored when the bird pecked the prey, carried it in the bill or otherwise touched the prey by the bill. Eating was scored when the bird ate at least part of the prey body. Cleaning was scored when the bird cleaned its bill by rubbing against the perch. Shaking was scored when the bird shook its body, close approaching was scored when the bird stood at the rotating tray and watched the prey from proximity (closer than 20 cm), far observation of the prey was scored when the bird looked at the prey from distant position (further than 20 cm).

Statistical analyses

Firstly, we performed multivariate analysis of all six recorded behaviours to 1) describe the correlation of the behaviours and 2) to estimate which behaviours are the most common for both bird species and particular prey types. We used Principle component analyses (PCA) in software Canoco 5.0 (Ter Braak and Šmilauer 2012) to visualize these relationships.

Subsequently, we ran generalized linear mixed effect model (GLMM, command glmer in R package lme4; Bates et al. 2015) to test the effect of the interaction of bird species and ladybird form on the occurrence of attack to the presented ladybird (coded binomially). As five ladybirds were offered to a single bird, the bird identity was included in the model as random factor. We used Likelihood ratio test for binomial data to compare the models in forward stepwise selection (Chi square test). Further, we used Fisher LSD post hoc test with Tukey correction for repeated comparisons to compare particular combinations of bird species and ladybird form.

We further analysed the edibility of particular ladybird forms, by analysis of eating the already attacked ladybird. We ran generalized linear mixed effect model (GLMM, command glmer in R package lme4) to test the effect of the interaction of bird species and ladybird form on the occurrence of eating (binomially coded, bird ID included as random factor). We used Likelihood ratio test for binomial data to compare the models in forward stepwise selection (Chi square test). Further, we used Fisher LSD post hoc test with Tukey correction for repeated comparisons to compare particular combinations of bird species and ladybird form.

Lastly, we analysed, how many of the five offered ladybirds each individual bird attacked to show any effect of aversion or habituation. We ran generalized linear model (GLM, command glm in R) to test the effect of the interaction of bird species and ladybird form on the number of attacked ladybirds by one birds (Poisson distribution of data). We used Likelihood ratio test for Poisson data to compare the models in forward stepwise selection (Chi square test). Further, we used post hoc test for Poisson distribution of data with Tukey correction for repeated comparisons to compare particular combinations of bird species and ladybird form.

All analyses were performed in R ver. 4.1.2 (R core Team 2022).