Skip to main content

Overlap in the wing shape of migratory, nomadic and sedentary grass parrots

Cite this dataset

Stojanovic, Dejan; Neeman, Teresa; Heinsohn, Robert (2020). Overlap in the wing shape of migratory, nomadic and sedentary grass parrots [Dataset]. Dryad.


Bird wing shape is highly correlated with mobility, and vagile species have more pointed wing tips than sedentary ones. Most studies of bird wing shape are biased to the northern hemisphere, and consider only two migratory syndromes (north-south migrants or sedentary species). There are major gaps in knowledge about the wing shapes of different taxa with other movement strategies (e.g. nomads) in the southern hemisphere. Parrots are a prominent southern hemisphere bird order with complex movement patterns, but their wing shapes are mostly unstudied. We test whether three metrics of wing shape of grass parrots (Neophema and Neopsephotus spp.) correspond to their purported migration syndromes (and other factors). We show that two strongly migratory grass parrots and an arid-adapted nomad had pointed wings, with flight feather longer distally and shorter proximally. However, purportedly sedentary species overlapped extensively with migrants and nomads in all aspects of wing shape, and taxonomic relationships, purported migratory syndromes and ecological barriers did not explain the variation we recorded. The most distantly related species (Neopsephotus) had most dissimilar wing shape to the others, but broadly conformed to the expectations of long pointed wings of a nomad. Why purportedly sedentary grass parrots had unexpectedly pointed wings is unclear. We propose the hypothesis that this wing shape may persist in sedentary populations if individuals experience strong but intermittent selection to disperse when environmental conditions are poor. If pointed wings are not costly during good times when individuals are sedentary, this wing shape may persist in populations as a ‘back up’ in bad times. Our study highlights the interesting migration patterns in the southern hemisphere that remain largely unstudied. Wing shape offers an interesting way to identify potentially undiscovered capacity for movement in data deficient species, which may also have implications for conservation.


We present data from 54 orange-bellied parrots, 46 blue-winged parrots, 30 elegant parrots, 31 rock parrots, 31 scarlet-chested parrots, 34 turquoise parrots and 32 Bourke’s parrots. There was a roughly even split between sexes in data for each species. We measured study skins at the Australian National Wildlife Collection, Australian Museum, American Museum of Natural History, Harvard Natural History Museum, Museum of Victoria, South Australian Museum and the Tasmanian Museum and Art Gallery. The mean collection date was 1946 (range: 1857 – 2016), and because museum skins stop shrinking after three years (Harris 1980, Green 1980) specimen age was unlikely to influence our study.  

Specimens were measured using electronic calipers (to the nearest 0.01mm) and a thin, soft plastic ruler (1 mm). We measured: (1) LW - unflattened wing chord, (2) unflattened length of the longest primary flight feather (measured from the point where the calamus inserted into the skin – we followed (Jenni and Winkler 1989) to measure the length of feathers), (3) ΔQ values (following the method of Lockwood et al. 1998, including the feather numbering system where p1 forms the leading edge of the wing), i.e. distances between the tip of each primary flight feather from the tip of the longest primary feather, and (4) SL - the distance between the carpal joint and the tip of the most distal secondary on the folded wing. These measurements are illustrated in Supplementary Materials Figure S1. We excluded juveniles (identified from specimen tags and metadata), specimens with broken or worn flight feathers, and specimens where the wings were not in the resting position. Because suitable wild-born specimens of all species (particularly orange-bellied parrots) were scarce in museum collections, we included some birds that had individual missing feathers (p4 – p8), and estimated the ΔQ value as midway between the two feathers adjacent the gap. We excluded birds missing p1 – p3. Either p9 or p10 were missing in 19 skins in our sample, so to minimize impacts on our sample size, we only included ΔQ1 – 8 values in the analysis and thus included these skins in the study. DS took all the measurements and quantified measurement error using a subset of repeat-measured birds. Measurement repeatability was high, and observer error accounted for mean 12.6 % of variance across the traits measured (range: 5 – 23%).  Additional information about the data and the species involved is provided in the paper.