Escalated begging does not compromise nestling health

A widely accepted explanation for the reliability of offspring begging signals assumes a differential benefits model balanced by direct viability costs independent of offspring nutritional condition. However, supporting evidence for this idea is inconclusive and often hampered by methodological limitations, including differential stimulation protocols and reliance on single, potentially biased markers of nestling health. This study tested the existence of direct, intrinsic, and condition-independent trade-offs between begging and body mass, immunity and oxidative stress by manipulating the begging effort of spotless starling (Sturnus unicolor) nestlings while maintaining constant food intake. We addressed potential problems of previous experimental protocols, ensuring uniform stimulation levels and evaluating multiple immune and oxidative markers. We observed no significant effects of experimentally increased begging effort in any of the 14 physiological markers analysed, with 95% confidence intervals of effect sizes consistently including zero or one (for the lysis capacity of plasma), indicating no biologically relevant effects. Overall, our findings suggest no physiological trade-offs associated with intense begging.

Methods
Experimental design

The study was conducted during the spring of 2022 in a nest-box breeding population of spotless starlings (Sturnus unicolor) in central Spain (Soto del Real, Madrid). Spotless starlings are medium-sized colonial passerines that typically lay two asynchronous clutches per season, with a modal brood size of four nestlings (Veiga and Polo 2016). Hatching dates were determined through regular inspections. All procedures were reviewed by the Animal Experimentation Ethics Committee of the University of Castilla-La Mancha and approved by the competent authority of the Junta de Comunidades de Castilla-La Mancha (approval reference: PR-2020-03-08), thus complying with current European, Spanish and Institutional guidelines for the care and use of animals.

The experiment involved 58 nestlings from 29 nests at the end of the linear growth phase (9-10 days after hatching, (Muriel et al. 2021). In the afternoon preceding the experiment (at 8 days of age), we selected two chicks of intermediate mass rank from each brood, ensuring at least two nestlings remained to prevent parental desertion. Initial blood samples (350 µl) were taken from the jugular vein of the focal chicks to measure basal levels of immune and oxidative markers. We also weighed the nestlings (accuracy = 0.1 g) and randomly assigned them to either a high-begging (HB) or low-begging (LB) treatment. Nestlings were then transported in a warm container to the laboratory, where they were placed in individual nest cups inside an incubator (RCOM 50 PRO, Autoelex, Korea) with controlled temperature (27-33 °C) and humidity (55-60%), ensuring identical environmental conditions and potential stressors. That afternoon, nestlings were conditioned to a standard begging stimulus (a playback of a parental nest-feeding call recorded from the same population) while being fed crickets (Acheta domestica) ad libitum.

Early the next morning (day 9), nestlings were weighed after the first meal of the day. We estimated the daily food intake based on their mass, following the allometric relationship by Weathers (Weathers 1996): daily food intake = 0.98 × M^0.814, where M is nestling body mass in grams. The daily food was divided into 12 equal portions, corresponding to 12 begging sessions conducted every hour from 8:00 to 20:00. Feeding intervals were based on mean intervals of parental feeding rates per nestling obtained from continuous recordings of 48 nests aged 6-8 days in our study population (mean interval ± SE = 59.05 ± 4.27 min). Any deviations from expected food intake were compensated for in subsequent sessions. The food consisted of alternating crickets and tiny omelette chunks (boiled egg), each weighed individually (Redondo et al. 2016). Whole egg is considered a high-quality diet (Moreno-Rueda and Redondo 2012), while insects are the bulk of the natural diet for nestlings (Veiga and Polo 2016).

During each feeding session, both nestlings in each dyad were carried together, covered by a cloth, to an adjacent room and stimulated to beg using the same playback of a feeding call. Each session included 1-4 trials in a random sequence determined beforehand as in Rodríguez-Gironés et al. (2001). Transporting both nestlings together ensured they experienced the same stressors and acoustic stimulation, which can influence glucocorticoid levels. To induce differences in begging effort, LB nestlings were fed immediately after gaping, while HB nestlings were stimulated to beg repeatedly by playing the playback and placing forceps over their mouths for 30 seconds longer than LB nestlings, resulting in a maximum cumulative duration of 2 minutes more per session in the 4-trial sessions. This duration was chosen to ensure differences in hourly begging rates between experimental groups comparable to previous studies assessing the physiological impact of begging. Our protocol resulted in differential begging rates higher than 45.5% of previous studies (Leonard et al. 2003; Moreno-Rueda and Redondo 2011, 2012; Moreno-Rueda et al. 2012; Redondo et al. 2016), exceeded by six studies (Soler et al. 1999; Kilner 2001; Rodríguez-Gironés et al. 2001; Moreno-Rueda 2010; Soler et al. 2014; Nettle et al. 2017).

Two 4-trial begging sessions (at 11:00h and 17:00h) were recorded with a digital video camera (Panasonic HC-V180) to validate the effects of the experimental manipulation on begging effort. A trained observer, blind to the experiment's purpose and chick treatments, analysed the videos at a reduced speed (0.125x) using Solomon Coder software (Version: beta 19.08.02, (Péter 2013). Postural begging intensity was assessed on a five-level ordinal scale: 0 (no begging), 1 (gaping, neck flexed, body horizontal), 2 (gaping, neck extended, body horizontal), 3 (gaping, neck extended, body off the nest floor, back reclining), and 4 (fully stretched, vertical back, sometimes wing flapping) (adapted from (Redondo and Castro 1992). For each trial, we recorded the time spent by the nestling at each postural intensity level and calculated the total duration of the begging bout by summing these durations. A summed begging score was computed by multiplying the duration of each begging level by its rank and summing them (Leech and Leonard 1996; Kilner 2001). The mean postural intensity was obtained by dividing the summed begging score by the total duration. Nestlings were weighed again at the end of the day.

The same procedure was repeated on day 10, except that after the last begging session, a 350 µl blood sample was collected from the jugular vein to analyse post-treatment immune and oxidative stress markers. All samples from both days 8 and 10 were collected while recording the time elapsed between opening the nest box or the incubator and sample collection (always < 3 min) to minimize disturbance effects on corticosterone (CORT) values (Jimeno et al. 2017). Samples were stored at 4°C until centrifugation (5 min at 4°C and 9,000g) within 2 hours after collection. Plasma and pellet were separated, the pellet was washed with cold saline solution, and both fractions were frozen at -80°C until analysis. Nestlings were returned to their original nests the following morning (day 11). At this moment, one randomly selected non-experimental nestmate was also blood-sampled within 3 minutes after opening the nest box to analyse potential differences in CORT levels due to laboratory procedures.

To verify that our experimental manipulation induced elevated begging in HB nestlings within the range of natural begging durations, wWe used data frommeasured individual durations of begging bouts in 164 feeding events of from 41 broods with 9-10 days-old chicks, recorded using an infrared video camera (BOBLOV PD50-32GB)., to calculate the duration of begging bouts under natural conditions. This allowed us to verify that our experimental manipulation induced elevated begging in HB nestlings, but still within the range of natural begging durations in wild broods. 

Immunological and oxidative stress assays

We quantified several immunological markers involving cell-mediated, humoral, and integrated immunity of both innate and acquired components. Specifically, we assessed the inflammatory T-cell mediated immune response elicited by an antigen using the phytohemagglutinin (PHA) skin test (Salvante 2006); the cellular capacity to fight infections and other diseases by quantifying the number of white blood cells (WBC) relative to 10,000 erythrocytes and the ratio of heterophils to lymphocytes (H:L ratio) in blood smears (Salvante 2006); complement activity and natural antibody levels using the lysis/agglutination assay (Matson et al. 2005); the acute-phase response to infection, inflammation, or trauma by measuring plasma haptoglobin levels (Millet et al. 2007); and plasma capacity to defeat micro-organisms by quantifying bacterial (E. coli) killing capacity (Matson et al. 2006).

We assessed oxidative stress by targeting different components of the system with several biomarkers of oxidative damage and antioxidant mechanisms (Halliwell and Gutteridge 2007; Monaghan et al. 2009; Pérez-Rodríguez 2009). Specifically, we measured a marker of oxidative damage by quantifying a by-product of lipid peroxidation in plasma (malondialdehyde acid, MDA) (Mateos and Bravo 2007); two markers of the combined action of nonenzymatic antioxidants in plasma, the Trolox Equivalent Antioxidant Capacity (TEAC) (Miller et al. 1993) and the OXY-adsorbent Assay (Costantini 2011); the absolute levels of a key intracellular endogenous antioxidant (total glutathione, tGSH) in erythrocytes; and an oxidative status index based on the functional levels of that antioxidant, the reduced to oxidized glutathione ratio (GSH:GSSG ratio) (Halliwell and Gutteridge 2007).

We quantified plasma CORT levels to ensure our stimulation protocol caused similar stress levels between both nestling groups and to control for potential effects of the levels of this hormone on begging and physiological variables. Detailed protocols for these techniques are available in the Supplementary material.

In our study, we rigorously followed a blind methodology for the PHA skin test, as measurements were made by a single person (S.C.) entirely unaware of nestling identity and previous laboratory procedures. Additionally, biochemical analyses were conducted blind to the experimental treatments, with samples from each dyad run in the same session in random order.

Statistical analyses

We performed statistical analyses using R 4.3.1. To enhance model stability, convergence likelihood, and accuracy of parameter estimates, we Z-transformed all continuous independent variables (mean-centered with an SD of 1) (Harrison et al. 2018).

To verify that our experimental manipulation (HB and LB treatment) created the expected differences among groups, we conducted separate linear mixed models with for time begging, begging score, mean postural intensity, and time spent at high postural intensity levels (4 and 5) as response variables, with the experimental treatment as the main predictor. We included the order of the begging trial (up to 4), the begging session (11:00 h or 17:00 h), and the day (day 9 or 10) as covariates, with the nest of origin nested within the date of arrival at the laboratory as random effects. We also assessed food consumption by nestlings using the total weight of food consumed (boiled egg and crickets) during the two-day experiment as the response variable. This model included the experimental treatment as the main predictor, initial body mass as a covariate, and the nest of origin nested within the date as random effects.

To assess the effect of the begging manipulation on body mass, corticosterone levels, immune function (PHA skin test, total WBC count, H:L ratio, lysis and agglutination capacity, haptoglobin levels, bacterial killing capacity), and oxidative stress markers (MDA, TEAC, OXY, tGSH, GSH:GSSG ratio), we used linear mixed models for each response variable. Lysis capacity was analysed using a generalized mixed model with a binomial error distribution due to low variability (76 samples had titres of 0, 28 of 1, and only 2 samples had titres of 2).

To control for initial levels of each physiological marker, we included the experimental treatment (HB or LB) in interaction with the time of the measurement (pre- or post-treatment) as the main predictor to analyse the change due to the treatment (Repeated measurements model). Initial mass (day 8) was included in all models as a covariate (except when mass was the response variable). All models included nestling ID nested within the nest of origin and the date of arrival at the laboratory as random effects. For the PHA skin test model, treatment was the only predictor since only one measure was taken per nestling (day 10), and nestling ID was therefore removed as a random effect. Additional covariates included handling time (time from opening the nesting box or incubator until blood collection) for the CORT model and levels of uric acid for the TEAC model (Cohen et al. 2007). Differences in MDA levels according to treatment and time were also checked, considering triglyceride levels as covariate (Pérez-Rodríguez et al. 2015).

To investigate potential time-dependent effects on mass gain (Soler et al. 2014; Redondo et al. 2016), we conducted separate models testing differences in body mass on the first (day 9) and last day of the experiment (day 10). In these models, the variable "time of measurement" was a categorical factor with two levels (beginning or end of the day). Failures to obtain sufficient blood volume in some individuals led to varying sample sizes between markers. We removed the entire dyad from the analysis of that marker if any measurement was missing (either initial or final values).

We confirmed the robustness of our results by means of two other additional analytical approaches. First, we replaced the categorical factor "Treatment" with a continuous variable accounting for the average time each nestling spent begging. Second, we used nestling dyads as our experimental unit and modelled how final HB-LB differences in physiological variables were explained by differences in begging as the main predictor (Within-Dyad differences model) (Kilner 2001; Leonard et al. 2003). Both approaches account for individual variability in begging effort within each "Treatment" level and inconsistencies in creating uniform differences in begging effort between HB and LB across all dyads (see Supplementary material for details).

To address multiple testing when analysing how experimental treatment affected multiple physiological variables, we applied a P-value correction using the Benjamini-Hochberg method to control for false discovery rate in begging, growth, immunological, and oxidative stress analyses (Benjamini and Hochberg 1995; Storey and Tibshirani 2003).

To measure the magnitude of the effects found in our study, we calculated effect sizes (adjusted standardized mean differences for dependent samples (Hedges' g), and odds ratio (OR) for the probability of lysis capacity) with their 95% confidence intervals (CIs), following recommendations by Nakagawa and Cluthill (2007) and Lakens (2013), using the esc R package (Lüdecke 2019). Additionally, we report standardized beta coefficients (β) with 95% CIs from our models to consider the influence of other factors (initial values, covariates, random effects) on the effect of the treatment on the physiological variables. These coefficients estimate standardized effect sizes for fixed effects in multiple regressions (Schielzeth 2010). We considered effect sizes whose CIs included zero (for Hedges' g or β) or one (for the odds ratio) as ‘no-effect’ (Nakagawa and Cuthill 2007; Tenny and Hoffman 2017).