Captive-breeding and release to the wild is a globally important conservation tool. However, captivity can result in phenotypic changes that incur post-release fitness costs, especially if they affect strenuous or risky behaviors. Bird wing shape is critical for migration success and suboptimal phenotypes are strongly selected against. I demonstrate surprising plasticity of bird wing phenotypes in captivity for 4/16 studied species. In a model species, captive-born juveniles with wild wing phenotypes (a 1mm longer distal primary flight feather) survived post-release at 2.7 times the rate of those with captive phenotypes (i.e. a shorter distal feather). Subtle phenotypic changes and their fitness impacts are more common than widely realized because they are easily overlooked. To improve captive-breeding for conservation, practitioners must surveil phenotypic changes and find ways to mitigate them.

There are two data sets included here: (i) shape_long.csv; (ii) ws.csv. Below i describe each in turn.

shape_long.csv:

I 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. I assigned individual provenance (captive/wild) based on specimen metadata. Captive specimen availability was patchy in museum collections – I aimed for at least five captive and wild specimens per species and excluded those that were under-represented. Species inclusion was limited by (i) collection bias toward attractive Australian native species which are preferred in captivity (Vall-llosera & Cassey 2017), (ii) for non-Australian species, absence of wild specimens for comparison, and (iii) sex-biased collections of some species (Cooper et al. 2019). Only specimens in perfect feather condition were included (e.g. molting individuals or those with broken feather tips were excluded). I selected common species in zoological and private collections because, like multi-generational conservation-focused captive-breeding programs, specimens were likely to be captive-born (not wild-collected). Although I previously showed that wing shape change in orange-bellied parrots (Neophema chrysogaster) is independent of generations of captive breeding, I aimed to minimize this risk in other species by using older captive-bred specimens that were less likely to be multi-generational captive-bred. However, the individual histories of captive-born specimens in this study were unknown. The mean collection date of captive specimens was 1955 vs. 1938 for wild specimens, reflecting the emergence of Australian avicultural trapping and trade last century (Franklin et al. 2014). In Figure 1 of the main paper, I illustrate the measurements taken in this study. Using electronic calipers (0.01mm) and rulers (1mm) I measured: wing chord (L_W), the length of the most distal secondary feather (L_S) and the length of the longest primary feather (L_P) (per Jenni & Winkler 1989). For the proximal and distal feathers adjacent to the L_P I measured the distance between their tips and the tip of LP (i.e. ΔQ, per Lockwood et al. 1998). The sample included 16 species from three families: Phasianidae, n=1 sp.; Psittaculidae, n=9 spp.; Estrildidae, n=6 spp. This wide array of species involves variance in wing shape; for some species, the L_P = P9 (i.e. second proximally from P10 on the leading edge of the wing) and for others, L_P = P8 (see Table S1). For simplicity, I hereafter refer only to the ‘proximal’ and ‘distal’ flight feathers adjacent the L_P (i.e. in some species this corresponds to P8 and P10 respectively, and in others it is P9 and P7). Furthermore, male princess parrots Polytelis alexandrae develop secondary sexual ornamentation – an elongated, spatulate tip on P8 – which makes this the L_P (Higgins 1999). In contrast, P9 is the L_P in female princess parrots and other Polytelis species (Higgins 1999). Consequently, I excluded male princess parrots to avoid skewing measurements due to sex. I aimed for equal sex ratios (Table S1) but due to collection biases (Cooper et al. 2019) and inclusion of monomorphic species without sex data, I did not always achieve parity.

ws.csv:

I measured 78 juvenile captive-bred orange-bellied parrots (Neophema chrysogaster) that were subsequently released to the wild over three years (2019: 30, 2020: 31, and 2021: 17). No wild-born parrots are included because at the nestling phase when wild birds are typically handled their wing feathers are still developing. In contrast, my sample of captive-bred parrots was between the ages of 3–5 weeks post-fledging, and their wing feathers were fully grown (Stojanovic et al. 2020a). Only juveniles with perfect wing feathers were included in this study. Juveniles are selected for release based on metapopulation management considerations (Morrison et al. 2020; Troy & Lawrence 2021) but not for any particular phenotypic trait bar good body mass and good general feather condition. Per Figure 1, I measured ΔQ of the feathers proximal (P8) and distal (P10) to the L_P (P9), which are known to vary between captive and wild parrots (Stojanovic et al. 2021). I dropped L_S but recorded L_W, tail length (L_T) and body mass (g). As an index of individual condition, I divided body mass by L_W to scale for overall size. I scored individuals as having survived (1) or died (0) their first year of life based on whether or not they returned from their first migration. I used the same criteria as previous survival analyses for this species using sightings data from supplementary feeders in the wild (Stojanovic et al. 2020d).