Symptoms of illness offer insights into an organism’s condition, altering social signals that affect others' behavior. Nestling birds employ begging signals to solicit parental care, but the extent to which begging reflects nestling health beyond hunger remains controversial. We investigated how experimentally induced changes in health affect begging signals in spotless starling (Sturnus unicolor) nestlings. To alter health status, we administered an immune challenge by injecting lipopolysaccharide (LPS) or a control substance (PBS) and observed begging behavior under controlled food deprivation conditions. LPS-treated nestlings exhibited delayed gaping, reduced stretching, and less begging time compared to control nestlings. Moreover, LPS-treated nestlings produced calls with higher spectral entropy, particularly when deprived of food for longer. Our findings indicate that begging signals are sensitive to induced sickness. However, addressing mechanistic issues is crucial for effectively testing whether begging reflects nestling conditions as a signal of quality in future experimental setups.

Experimental design

We conducted the study in a nest-box breeding population of spotless starlings located in central Spain (Soto del Real, Madrid). In our study population, spotless starlings, medium-sized colonial passerines, typically have two asynchronous clutches per season, with a modal clutch size of 5 eggs (López-Rull et al. 2007; Salaberria et al. 2014). We determined the hatching date of the nests through regular inspections.

We chose 25 nests from the first wave of clutches in the breeding season. On the day before the experiment, at the age of 5 days, we selected two chicks of intermediate-mass rank from each brood, ensuring that at least two nestlings remained to prevent parental desertion. The nestlings were transported in a warm container to the laboratory, where they were placed together in artificial nest cups with controlled temperature (approximately 35°C) and humidity (45%). The nestlings were familiarized with a begging stimulus, which was a playback of a parental feeding call recorded from the same population while being provided ad libitum with super worms (Zophobas morio).

On the following morning (day 6 of age), we weighed the nestlings with a digital balance (accuracy = 0.1 g), placed them in individual nest cups side by side with the nest cup of their sibling, and provided them with three different meals to ensure a uniform nutritional condition. In the first two meals, nestlings had ad libitum access to superworms and crickets (Acheta domestica), respectively, with a 45-minute gap between feedings, during which we recorded the number of insect items consumed. These intervals were determined based on the mean intervals of parental feeding rates obtained from a random video sample of 48 nests in our study population (mean interval = 59.05 min ± 4.27 SE; our data). We calculated the average weight of each superworm and cricket by weighing 10 pools of 10 units of each insect type. On average, one superworm larva weighed 0.5 g, and one cricket weighed 0.14 g. In order to estimate the total amount of food consumed by nestlings, we multiplied these weights by the number of insects of each type consumed per nestling. After 45 min, we offered nestlings high-quality food (i.e. boiled egg, Moreno-Rueda and Redondo 2012) in an amount equivalent to 3.5% of their body mass. This food quantity was estimated using the allometric relationship from Weathers (1996): daily food intake = 0.98 M^0.814, where M represents nestling body mass in grams, divided into 14 daily portions. Egg quantity was also weighed to estimate the total food consumed by nestlings. Following the last meal, nestlings were individually transported to a recording chamber to monitor their begging behavior. This first test was used to determine baseline begging behavior. Begging behavior was recorded at 30, 60, and 90 min after the last meal. During each begging trial, nestlings were stimulated twice using the same stimulus as before (feeding call playback), allowing begging to cease completely between both stimuli. Postural begging was recorded using a digital video camera (Panasonic HC-V180), and acoustic begging was recorded using a digital audio recorder (Marantz Professional PMD620) placed 5 cm from the nest cup. After the final begging trial, nestlings were provided ad libitum with super worms, and siblings of each dyad were randomly assigned to either an LPS or control treatment. Nestlings in the LPS treatment received an intraperitoneal injection of 50 μl of LPS (ref: L2880, Sigma-Aldrich) suspended in a phosphate-buffered saline solution (PBS) at a concentration of 1 mg/kg body weight. Control nestlings received the same volume of PBS alone. This dosage was selected based on the dose-dependent results reported by Armour et al. (2020) in the northern bobwhite quail (Colinus virginianus) and was the same dose used by Serra et al. (2012) in European Starling (Sturnus vulgaris) nestlings.

1h and 15 min later, we initiated feeding to equalize the short-term food deficit, following the same protocol described above, with feeding every 45 min. This ensured that the second begging test occurred 3-4.5 h after the injection, allowing the chicks to remain influenced by LPS while avoiding the peak effect of the challenge (for justification, see below). We deliberately chose not to conduct the post-treatment tests at the peak effect of the challenge (1-2 hours after the injection) as we anticipated that, at that point, nestlings might be in a weakened state, potentially yielding unreliable responses (see below). Post-injection test was carried out in the same manner as the initial one, recording the behavior of the nestlings at 30, 60, and 90 min following the last meal (i.e., at 3 h 15 min, 3 h 45 min, and 4 h 15 min after the LPS/Control injection). This approach allowed us to observe changes in the behavior of a focal nestling in response to both LPS and Control treatments, thereby testing both within- and between-subject effects. At the conclusion of the experiment, we feed nestlings again with super worms recording the number of insects to estimate the amount of food consumed as before, we weighed nestlings, collected a blood sample (approximately 50 µl) from the brachial vein for genetic sexing, and returned them to their original nests by the end of the day. Nestlings were weighed again on day 8 of age (i.e., 48 hours after the laboratory experiment ended) to detect any potential medium-term effects of the experimental treatment on their mass.

The timing of the post-injection begging test was synchronized with the effect of LPS on nestling temperature, based on a prior pilot test involving 17 nestlings of the same age as those used in the experiment (see Supplementary Material for additional details). In this pilot assay, we subjected nestlings to the same LPS injection protocol and doses mentioned previously and measured variations in cloacal temperature. The most pronounced effects of LPS on temperature were observed within the initial 3 hours following the injection, with the peak temperature difference between LPS and Control nestlings occurring at 2 hours post-injection (see Supplementary Figure S1). These results were consistent with previous studies employing similar injection protocols (Owen-Ashley et al. 2006; Burness et al. 2010), suggesting that the peak effects of LPS on body temperature, interleukin-mediated inflammation, and corticosterone levels (Nakamura et al. 1998; De Boever et al. 2008; Burness et al. 2010) manifest within 1-2 hours after injection.

Analysis of postural and acoustic traits

An external trained observer, blind to the experiment's purpose and chick treatments, analyzed videos at a reduced speed (0.125x) using the Solomon coder software (Version: beta 19.08.02; Péter 2013). Postural begging intensity was assessed on a five-level ordinal scale, indicating the degree of body stretching: 0 (no begging), 1 (gaping, lying body), 2 (gaping, neck extended, lying body), 3 (neck extended, body separated from nest floor, back reclining), and 4 (gaping fully stretched, vertical back, sometimes including wing flapping) (adapted from Redondo and Castro 1992). At each sampling time, we recorded the time spent by the nestling at each postural intensity level and calculated the total duration of the begging display by summing the durations of each posture. Additionally, we computed the mean and maximum postural ranks for the whole begging display. The mean postural intensity was computed by multiplying the duration of each begging level by its rank and then dividing it by the total duration (Leech and Leonard 1996; Kilner 2001). We also recorded the latency time from the start of each stimulus to the onset of the begging (precision = 0.1 s).

We used Avisoft SAS-Lab Pro v. 5.1.23 software (Specht 2002) to analyze begging vocalizations. Spectrograms were created with the following settings: 256 FFT, FlatTop window, frame 100%, overlap 50%, a bandwidth of 162 Hz, and a high-pass filter at 2 kHz, resulting in a time resolution of 11.6 ms and a frequency resolution of 43 Hz. From these spectrograms, we extracted six temporal and spectral measurements commonly examined in nestling begging behavior studies that are related to offspring cryptic need (hunger) and that are known to vary with modifications of their internal state (e.g. Sacchi et al. 2002; Reers and Jacot 2011; Klenova 2015). We manually measured the total vocal bout duration (in ms) from the start of the first call to the end of the last one and the total number of calls. The calling rate was calculated by dividing the total number of calls by the bout duration. We selected a subsample of the first 10 calls in each bout following each playback stimulus to obtain spectral measurements. For each begging bout, we measured call duration, peak frequency, maximum and minimum frequency, call entropy (energy dispersion across the frequency spectrum), and the lower (25%) and upper quartiles (75%) of energy distribution across the frequency spectrum. Spectral variables were measured at the point of maximum amplitude of the element, both instantaneously and after specifying a minimum hold duration of 40 ms (peak hold mode). We also obtained measures for spectral variables considering either the Max or Mean spectrum of the entire element. Due to a recording system failure, no acoustic measurements were obtained in the two nestling dyads.

Statistical analysis

We performed statistical analyses using R 4.1.3 (R Core Team 2022). To enhance model stability, the likelihood of model convergence, and the accuracy of parameter estimates, all independent variables were Z‐transformed (mean-centered with an SD of 1) (Harrison et al. 2018).

To assess the impact of the experimental treatment on nestling conditions, we examined short- and medium-term effects on body mass. We used a linear mixed model to analyze differences in body mass at the end of the experiment (day 6), controlling for the initial mass (day 5). Additionally, we conducted two more analyses to explore medium-term effects on body mass when chicks returned to their nests after the laboratory procedures. For this purpose, we used a model with body mass on day 8 (two days after returning to the nest) as the response variable, controlling for the initial mass (day 5), and a separate model controlling for the mass at the end of the laboratory procedures (day 6). In the last two models, we included brood size as a covariate to account for differences in individual feeding rates due to sibling competition. Experimental treatment (LPS or Control) served as the main predictor in all these models. To evaluate the effect of treatment on nestling food intake, we used the total amount (g) of food consumed before each test period as the response variable. In this model, we included experimental treatment (LPS or Control) interacting with the test period (before or after the injection) as the main predictor and the initial body mass as a covariate. The nutritional values, chemical and chitin compositions, digestibility, and handling times differed among the three food types (superworms, crickets, and boiled egg) fed to nestlings (Kaspari and Joern 1993; Banbura et al. 1999; Finke 2002; Moreno-Rueda and Redondo 2012; Oonincx and Finke 2021). As this may affect nestling predisposition to feed and their mass gain, we conducted three additional models with each food type separately as the response variables. We also conducted an additional analysis to evaluate the effect of LPS on the amount of superworms consumed at the end of the experiment (after the second begging test).

To analyze the effects of LPS treatment on postural begging, we employed linear mixed models with latency time to gape, mean begging intensity, and duration of the begging display as response variables. The maximum postural intensity was examined using an ordinal mixed model (Christensen 2022). In all these models we included the experimental treatment interacting with the test period as the main effect. To account for the influence of hunger, we included time of food deprivation (30, 60, or 90 min since the last feeding) as a three-level factor covariate in a three-way interaction with treatment and test period. If this interaction proved non-significant, it was excluded from the final model. We incorporated initial body mass and order of stimulus (1 or 2) as covariates. For analyses of acoustic begging, we applied the same approach as described above to assess calling rate and call duration. We also conducted individual Principal Components Analysis (PCA) for peak frequencies, maximum frequencies (including quartile 75%), minimum frequencies (including quartile 25%), and entropies. This allowed us to combine various measurements obtained from different sections of a call, including the point of peak amplitude with and without peak hold mode, as well as the Max and Mean spectra of the entire element (Supplementary Table S2). We used the scores on the first component (PC1) as response variables and analyzed them using the same approach. Additionally, we included the sequential order of calls as a covariate since we analyzed up to 10 consecutive calls per playback stimulus. The effect of nestling sex was examined in all models as a covariate; however, it did not significantly influence any of the response variables and was therefore omitted from the final models.

We determined the optimal random-effect structure for each of the aforementioned models using a top-down strategy. Nested models were compared through likelihood ratio tests (Zuur et al. 2009). We included chick nested within a nest of origin and both nested within the date of the experiment as random intercept effects.  Experimental treatment, test period, and time of food deprivation (considering their effects alone, additive, or interacting) were included as random slope effects.

To address the issue of multiple-testing with the experimental treatment and time of food deprivation predicting various begging variables, we incorporated a P-value correction (Benjamini-Hochberg method) for postural and acoustic analyses separately, to control for false discovery rate in the statistical tests (Benjamini and Hochberg 1995; Storey and Tibshirani 2003).