This dataset provides the raw data collected from wear and tensile tests. In the wear assay, 24 melanized and 24 non-melanized pigeon first primary feathers were tested, and the lost area (%) of each feather was measured. In the tensile assay, 18 barbs from each of the 4 groups - unbleached melanized, unbleached non-melanized, bleached melanized, and bleached non-melanized feathers - were tested. The measured parameters were as follows: breaking force, work to break, cross-sectional area, breaking stress, breaking strain, tensile modulus, and second moment of inertia. All values were ready for subsequent statistical analysis.

Wear assay:

Feather wear was tested by impacting/striking their ventral sides against four obstacles equipped with the same substrate. All the feathers were the first primary feathers located in the middle of a wing, with generally the same size, shape, and length. A pair of feathers, one melanized and one non-melanized, was first fixed on a centrifuge using a 3D-printed clamp. Subsequently, their ventral sides were subjected to impact against threaded rods wrapped with abrasive discs of 12 µm grain size. The threaded rods were evenly fixed around the periphery of the centrifuge, allowing four impacts per cycle. The tested feathers were spun at 500 rpm for 25 min. Only the vane tip, at a specific distance h from the distal end, made contact with abrasive discs during the wear assay. This particular area is considered the total area of the feather for comparison with the lost area. The percentage of the lost area was calculated by dividing the actual lost area A1 by the total area A1+A2. The images of feathers, captured before and after the wear assay, were taken using the 3D-microscope Keyence VR-3100, and the lost areas of the barbs were measured using ImageJ 1.53k. A total of 24 melanized and 24 non-melanized feathers were tested.

Tensile testing:

The mechanical properties of feather barbs were tested by tensile measurements. A single barb was first cut off from the distal region of the feather vane and fixed between two pieces of postcard paper using the 5925 Ergo® fast glue. Subsequently, the attached pieces of cardboard were fixed in the clamps and stretched vertically at a speed of 0.1 mm s^-1 using a Zwick static material testing machine until the barb broke. The force-displacement curve of each measurement was recorded by the software Zwick analyzer testXpert II v.3.5. Simultaneously, video monitoring of the tensile process was performed using a digital camera. Both melanized/non-melanized and untreated/bleached barbs were tested. A total of 24 untreated melanized barbs, 38 untreated non-melanized barbs, 43 bleached melanized barbs, and 19 bleached non-melanized barbs were tested. All barbs were collected from the same region near the tip of the first primary feathers from 52 individuals. The unequal initial sample numbers were due to additional tensile tests conducted to obtain sufficient specimens with fracture surfaces suitable for SEM-based cross-sectional measurements. The final dataset used for statistical analysis was balanced (n = 18 per group).

For the calculation of tensile parameters, breaking force was defined as the maximum force recorded during the tensile test. Work to break was calculated as the area under the force-displacement curve. After the fracture, SEM images of the fracture surface were obtained to determine the cortex cross-sectional area. The stiffness of the medulla was quantified as 1% of the stiffness of the cortex (Bonser 1996). Therefore, here we chose the cortex area to represent the barb cross-sectional area (A) for comparisons and subsequent mechanical property calculations. The second moment of inertia (I) was estimated using the geometric approximation I = A²/5.027, where 5.027 corresponds to the elliptical correction factor for a 2.5:1 aspect ratio (major axis: minor axis) determined by the general elongated shape of multiple cross-sections of the barbs. The normalized breaking stress and strain were calculated as the breaking force divided by the cross-sectional area and the elongation during the tensile process divided by the original length of the barb, respectively. The tensile modulus was calculated as the stress divided by the strain. Since accurate determination of the cross-sectional area at the tensile fracture site is critical, only barbs that fractured cleanly and could be positioned vertically without surface contamination (e.g., adhesive coverage) were included in the cross-sectional measurements and in the other calculations.