Morphology is closely linked to locomotion and diet in animals. In animals that undertake long-distance migrations, limb-morphology is under selection to maximize mobility and minimize energy expenditure. Migratory behaviors also interact with diet, such that migratory animals tend to be dietary generalists, while sedentary taxa tend to be dietary specialists. Despite a hypothesized link between migration status and morphology, phylogenetic comparative studies have yielded conflicting findings. We tested for evolutionary associations between migratory status and limb and bill morphology across kingbirds, a pan-American genus of birds with migratory, partially migratory, and sedentary taxa. Migratory kingbirds had longer wings, in agreement with expectations if selection favors improved aerodynamics for long-distance migration. We also found an association between migratory status and bill shape, such that more migratory taxa had wider, deeper, and shorter bills compared to sedentary taxa. However, there was no difference in intraspecific morphological variation among migrants, partial migrants, and residents, suggesting that dietary specialization has evolved independently of migration strategy. The evolutionary links between migration, diet, and morphology in kingbirds uncovered here further strengthen ecomorphological associations that underlie long-distance seasonal movements in animals.

We measured bill length, width, and depth at the distal end of the nares. We also measured unflattened wing chord length, Kipp’s distance (a measure of wing pointedness: the distance between the tip of the first secondary feather to the tip of the longest primary feather; Kipp, 1942, 1958; Baldwin et al., 2010), tail length, and tarsus length on 2108 study skins from across the ranges of each species and subspecies (28 operational taxonomic units (OTUs)). MM measured 2008 specimens, and identified all individuals to the lowest level of taxonomic identification (using Clements et al., 2019) and classified each individual as migratory, partially migratory, or sedentary (sensu Fitzpatrick et al., 2004). JIGA measured 100 specimens to improve sampling of some taxa. Bill and tail measurements were taken twice and were confirmed to be within one mm of one another. Both right and left wing chords and tarsus lengths were measured and averaged. Measurements by JIGA were only taken once. We measured tail length as the longest rectrix to the nearest 0.1 cm (Pyle et al., 1997). For most measurements, we used a Mitutoyo brand IP 67 digital calipers (part number 573-271) with a range of up to 15.24 cm, with 0.00127 cm resolution. For tails longer than 15.24 cm, we used a 30.48 cm stainless steel ruler placed between the two middle rectrices. When tails were longer than 30.48 cm photos were taken of the tails above a 0.64 x 0.64 cm square grid with the calipers measuring to their extent and ImageJ was used to calculate the full tail length (Supporting Information 2). We then averaged measurements across individuals within each OTU for downstream analyses.

The readme file contains an explanation of each of the variables in the dataset, and its measurements units. Information on how the measurements were done can be found in the associated manuscript referenced above. Below we detail special cases to be aware of in this dataset.

Some specimen tags did not reflect complete data, and when data was incomplete we made a couple of adjustments that are reflected in the attached data set. These did not affect our analyses, but should be noted for future use of the dataset. For specimens that were dated to 18--, meaning some time in the 1800s, we changed the cell to read 'null'. This occured for the following specimen numbers that all originated from the Smithsonian collection: 120074, 52806, 7389, 38191, 39457, 49586, 71196, 101425, 129146. Some specimens had collection years listed without the century noted (e.g., '40 instead of clarifying whether the specimen was from 1840 or 1940). We assumed these were from the 1900s because that was most likely and changed the year to 1940 in the year column. This occurred for the following specimen numbers that all originated from the Smithsonian collection: 356917, 356918, 359886, 531616, 531617. For specimen 30633 from the Louisiana State University Museum of Natural Science collection, the exact date was not specified and instead a potential range from 20-25th of the month. For this specimen we selected the first possible day (i.e., the 20th in this case) for the day column. Lastly, two specimens from the Royal Ontario Museum did not have catalogue numbers, but instead had an older version of a collection number (i.e., "30,6,21,450" and "33,9,1,676"). We changed these to numbers without commas to match how the specimen ID was labelled for the rest of the dataset. For any cells where no data was given on specimen tags for that specimen in any column, the cells read 'null'.

Accurate morphological measurements were not possible for all metrics used in this study, and where these were not possible the cell reads 'null'.