Carotenoid-based plumage color is crucial in avian mate selection, often serving as an indicator of individual quality. To determine whether carotenoid-derived color can be a sign of individual condition and if there is a relationship between an individual’s condition and color production, it is necessary to identify how carotenoids are acquired by individuals and subsequently used by the organism. Our objective was to determine how carotenoid pigments are used in the stripe-tailed yellow finch Sicalis citrina, a species wherein females exhibit a light yellowish ventral color while males are bright yellow. By using carbon and nitrogen stable isotope analyses we were able to determine if these carotenoid-derived colors are a signal of individual condition in terms of physiological stress and body condition. Females with low δmode_comment¹³C values (≈ −18‰) indicating the consumption of C₃ and C₄ seeds, exhibited colors shifted toward longer wavelengths and better body condition, and those with such color shifts also had higher carotenoid concentrations. In contrast, brighter females had higher δ¹⁵N values, indicating greater consumption of arthropods. Males with more saturated ventral patches had higher carotenoid concentrations, while those with colors shifted toward shorter wavelengths or lower H/L ratios, i.e. less stress, exhibited high δ¹⁵N values, suggesting that they supplement their diet with arthropods (δ¹⁵N ≈ 5‰). Our results show that the carotenoids dynamics in stripe-tailed yellow finches differ between sexes and highlight how food sources impact condition, stress, and ornamentation. Our study indicates that sexual dimorphism extends beyond plumage color to deeper physiological and ecological differences.

We carried out fieldwork to capture stripe-tailed yellow finches in the Brasília National Park during two breeding seasons: 2018/2019, and 2019/2020. We identified seven sites of greatest activity of individuals and set up three to five mist nets in the morning, totaling a minimum capture effort of 2268 h.m² per site. To maximize capture success, we changed the locations of mist nets daily, setting up the mist nets in the territories of the males we heard singing and/or displaying. We captured about 1 to 3 individuals at a time and as soon as they were captured, we (who were in hiding) removed them from mist nests. We banded all captured individuals with metal bands provided by CEMAVE (Brazilian National Center for Research and Conservation of Wild Birds), and a unique combination of colored plastic bands. We identified females as those individuals with incubation patches and males by their complete yellow plumages, sampling only these males since our focus is on carotenoid pigments. We collected five to ten ventral feathers to determine their colors and carotenoid type. We chose ventral feathers because this is the part of the body has the most saturated yellow color and which is most visible to other individuals during displays. Also, the ventral part is often one of the most important in social signaling (Shultz and Burns 2017). Feathers were attached by their calamus with tape to pieces of cardboard and wrapped with aluminum foil for protection from light and moisture. We also collected blood samples through brachial vein puncture, using a 0.45 × 13 mm needle and non-heparinized capillary tubes. To assess physiological stress (described below), we used a drop of blood to produce blood smears on slides, which were fixed and stained with the rapid panopticon kit Instant Prov. We ensured that blood samples were collected immediately after biometric measurements and before any feather collection, minimizing the impact of handling stress on the analyses.

To estimate the body condition, we measured tarsal length with Digimess® digital calipers (300 mm) and weighed all individuals with a precision spring balance (30 g). Finally, we collected one wing feather (first primary remige) to determine the relationship between diet, using the stable isotope protocol described below, and carotenoids color expression. These samples were stored in small paper bags. We handled all individuals for a maximum of five minutes to minimize their stress.

Ornament color and carotenoids

First, we measured the ventral color of all individuals using the spectrophotometer Ocean Optics® USB4000 with pulsating xenon light source PX-2 (Ocean Optics, Dunedin, Florida), which enables reflectance in the ultraviolet spectrum. We obtained the measurements by positioning a single optical fiber at a 45° angle to the feathers and collected the backscattered radiation along the same axis (Pellegrino et al. 2020). We placed feather samples on a black, spectrally flat surface and recorded the average of three spectral measures (to the right of the rachis, above, and the left) using the OceanView® software from Ocean Optics (https://www.oceanoptics.com/software/oceanview/.). The spectra were interpolated at 1 nm intervals over the range of 300–700 nm. We used the R package ‘pavo’ (www.r-project.org, Maia et al. 2019) to smooth spectral curves and quantify the brightness, hue, and feather saturation. For this, we applied visual modeling considering the Blue Tit Cyanistes caeruleus visual system, and set the models assuming illuminant conditions for open habitats, where the stripe-tailed yellow finch breeds. We calculated brightness (‘luminance’), a metric related to the structural quality of the feathers, as the intensity of reflectance across the entire spectral range measured (300–700 nm). Saturation (‘r-achieved’), a more relevant metric for analyzing carotenoid concentration, was quantified as the relative distance of a given color (hue) from the achromatic origin (Stoddard and Prum 2008, Demko et al. 2020). Lastly, we determined the hue in the red-green-blue (RGB) and ultraviolet (UV) spectra (θ and Φ angles in radians). Higher RGB and UV hue values indicate a shift toward shorter wavelengths. Only hue in RGB and UV spectrum were highly correlated for both sexes (r_p > 0.70, p < 0.005;', Supporting information), therefore, hue in UV was disregarded from further analyses. We chose to keep the hue in RGB measurement because carotenoids are primarily expressed in this spectrum.

Second, we used these same ventral samples to identify carotenoid types and estimated their concentration using ventral feathers of females and males with complete yellow plumage. We analyzed these samples in the Toxinology Lab of the University of Brasília, Brazil. We weighed samples on an analytical balance (0.0001 g) and placed them in Falcon tubes (15 ml) containing 3 ml of a solution composed of acetone:petroleum ether:didethyl ether (1:1:1). Subsequently, we subjected these samples to an cell disruptor for 3 cycles of 2 minutes each with an interval of 1 minute. After 48 hours at 5°C, each tube was centrifuged at 12 000 RPM for 10 minutes. Then, we removed the solution and placed it in a new Falcon tube, which was subjected to vacuum drying in a speedvac system. After drying, the carotenes were resuspended in 400 μl of methanol and filtered through a 0.45 μm membrane for analysis in a high-performance chromatography system (HPLC) LC-10 (Shimadzu, Japan) with the analytical column C-18 KINETEX Phenomenex (0.,45 × 25 cm), oven at 35°C, mobile phase 100% methanol, and a photodiode detector in the range of 300–600 nm. We identified the carotenoids present in the feather samples according to the comparison with the lutein, alpha-carotene, and beta-carotene standards of retention time and absorbance spectrum of 300–600 nm.

We quantified the carotenoid concentration most well represented in the samples (lutein) by comparison with carotenoids that occur in cabbage. For this, we ground 50 g of cabbage and extracted the carotenoids separately with a 50 ml solution of acetone:petroleum ether:diethyl ether (1:1:1). The extract was divided into two 50 ml Falcon tubes and subjected to the cell disruptor for 3 cycles of 2 minutes each with an interval of 1 minute. After 48 hours at 5°C, tubes were centrifuged at 12 000 RPM for 10 minutes. We removed the extracts and conditioned them in new Falcon tubes (15 ml) which were then submitted to vacuum drying in a speedvac system. After drying, the carotenes were resuspended in 1 ml of methanol and filtered with a 0.45 μm membrane for analysis in an HPLC system. We collected and dried the lutein carotenoids separately in a vacuum system. These carotenoids were quantified according to Britton (1995).

Physiological stress

The blood smears were used to assess the physiological stress of individuals. White blood cell counts (heterophils, lymphocytes, monocytes, eosinophils, and basophils) were conducted by the commercial Santé Veterinary Lab, Brasília, DF. White blood cells, as part of the body’s immune defense system (Miesle 2011), have been used to study health parameters in several taxa (Dehnhard et al. 2011, Simons et al. 2012). Given that in the exploratory phase of our analyses, we found no eosinophils or basophils, that the number of monocytes was close to the normal range (0–3%) (as suggested by Miesle 2011), and that the numbers of heterophils, lymphocytes, and the heterophil/lymphocyte (H/L) ratio were highly correlated in females and males (r > |0.95|; p < 0.001), we chose to assess the levels of stress of individuals using the H/L ratio. The H/L ratio has been established as a robust measurement of physiological stress, such that an elevated H/L ratio implies a higher level of stress (Gross and Siegel 1983, Davis et al. 2008, Davis and Maney 2018). The H/L ratio changes in response to external stressors that typically represent the risk of injury to individuals, which may include weather changes, social challenges, increased reproductive effort, or parasite infestation (Minias et al. 2019).

Body condition

Body condition results from accumulated energy in the individual's body as a result of food ingestion and can be an indicator of individual condition (Peig and Green 2009). A body condition index (BCI) was calculated for all individuals using the scaled mass index, due to its potential to remove covariance between body size and other body components (e.g. tarsal size) (Peig and Green 2009). We used the calculation provided by Peig and Green (2009):.

Stable isotope analysis

We used the nitrogen (δ¹⁵N) and carbon (δ¹³C) isotope ratio to estimate the food sources from which carotenoid pigments were extracted. Despite the high seasonality and dry conditions in the study area, which can limit plant transpiration and photosynthesis, leading to enrichments in ¹⁵N and ¹³C (Shipley and Matich, 2020), our systematic data collection (e.g. using the same feather sample) controls for potential isotopic differences due to water stress and maintains result consistency. Furthermore, our interpretations are grounded in an understanding of Cerrado dynamics and its isotopic fractionation, as detailed below.

As δ¹⁵N can be used to estimate trophic position, it may indicate the consumption of prey and assimilation of animal protein, with the smallest values indicating the consumption of seeds directly and the highest values suggesting prey consumption, in our specific system. In the Cerrado, the δ¹⁵N basal of open areas with native and exotic grass vegetation is on average −1.70‰ and −1.86‰, respectively (Sena-Souza et al. 2023). Accounting for isotopic fractionation in the tissue, δ¹⁵N close to these values indicates that individuals occupy a low position in the trophic chain, i.e. primary consumers. On the other hand, lower δ¹³C values (average in −21‰) suggest that an individual´s food sources are Cerrado native C₃ grasses, while higher δ¹³C values (average in −14‰) suggest that the food sources are native or exotic C₄ grasses, such as molasses grass (Sena-Souza et al. 2023, Navarro et al. 2023). Intermediate values suggest a mixed C₃ and C₄ diet (Zhao et al. 2010, Andriollo et al. 2017, Navarro et al. 2023, Sena-Souza et al. 2023).

We collected the first primary feather of both females and males. These feather samples reveal the isotopic values corresponding to the beginning of the breeding season of stripe-tailed yellow finches, since they molt prior to breeding (unpublished data, de-Carvalho unpubl.). We cleaned these samples by leaving them submerged for 30 minutes in a 2:1 solution of chloroform and methanol to remove dust particles and oil residues (Paritte and Kelly 2009). After this step, feathers were oven-dried at 50°C for 48 hours, the distal part of the vane was cut, weighed (~ 0.8–2.0 mg), encapsulated in tin capsules, and sent for isotope-ratio mass spectrometry analysis. The nitrogen and carbon isotope ratios were determined by combustion using an elemental analyzer (Carlo Erba, CHN-1100) coupled to a Thermo Finnigan Delta Plus mass spectrometer at the Laboratory of Isotope Ecology of the Centro de Energia Nuclear na Agricultura (CENA/Universidade de São Paulo), Piracicaba, SP, Brazil. The results are expressed in delta notation (δ), in parts per thousand (‰), based on internationally recognized standards. We used the following equation: δ¹⁵N or δ¹³C (‰) = (Rsample – Rstandard)/Rstandard × 1.000), where Rsample and Rstandard represent the heavy/light isotope molar ratio of the sample and standard, respectively. The standard used for nitrogen analysis was atmospheric air (¹⁵N:¹⁴N ratio = 0.0036765) and the standard used for carbon analysis was Vienna Pee Dee Belemite (Vienna PDB; ¹³C:¹²C ratio = 0.01118). Internal standards (tropical soil and sugarcane leaves) are routinely interspersed with target samples during analyses. The Llong-term analytical error for the internal standards is 0.2‰ for both δ¹⁵N and δ¹³C.