Effect of captivity on the vertebral bone microstructure of Xenarthran mammals

Captive specimens in museum collections facilitate study of rare taxa, but the lifestyles, diets, and lifespans of captive animals differ from their wild counter- parts. Trabecular bone architecture adapts to in vivo forces, and may reflect interspecific variation in ecology and behavior as well as intraspecific variation between captive and wild specimens. We compared trunk vertebrae bone microstructure in captive and wild xenarthran mammals to test the effects of ecology and captivity. We collected μCT scans of the last six presacral vertebrae in 13 fossorial, terrestrial, and suspensorial xenarthran species (body mass: 120 g to 35 kg). For each vertebra, we measured centrum length; bone volume fraction (BV.TV); trabecular number and mean thickness (Tb.Th); global com- pactness (GC); cross-sectional area; mean intercept length; star length distribu- tion; and connectivity and connectivity density. Wild specimens have more robust trabeculae, but this varies with species, ecology, and pathology. Wild specimens of fossorial taxa (Dasypus) have more robust trabeculae than cap- tives, but there is no clear difference in bone microstructure between wild and captive specimens of suspensorial taxa (Bradypus, Choloepus), suggesting that locomotor ecology influences the degree to which captivity affects bone micro- structure. Captive Tamandua and Myrmecophaga have higher BV.TV, Tb.Th, and GC than their wild counterparts due to captivity-caused bone pathologies. Our results add to the understanding of variation in mammalian bone micro- structure, suggest caution when including captive specimens in bone micro- structure research, and indicate the need to better replicate the habitats, diets, and behavior of animals in captivity.

Methods
We chose 44 adult xenarthran specimens from the FMNH Mammalogy collections to undergo X-ray micro-computed tomography (μCT) scanning at the University of Chicago’s PaleoCT facility (https://luo-lab.uchicago.edu/paleoCT.html). We scanned the last six presacral vertebrae of each specimen. We scanned all specimens using the 240 kV tube of the PaleoCT lab’s GE v|tome|x μCT scanner at resolutions from 22.101 µm to 96.551 µm (average resolution 44.899 µm), depending on specimen size. We reconstructed the scans in GE phoenix datos|x and aligned and cropped the reconstructed scans using VGstudio (Volume Graphics, 2019). We segmented prism-shaped volumes of interest (VOI) from the vertebral centra. We determined the size and location of the prism by orienting the vertebra in cranial view and selecting the largest 2D square area of trabecular bone at the dorsoventrally and mediolaterally narrowest point of the centrum. Using the “Threshold” tool in FIJI, we converted the VOI into a binary image stack in which we maximized bone sampling and minimized sampling of non-bone materials. We additionally segmented out entire vertebrae including only the slices where the vertebral foramen was completely surrounded by bone and binarized them using the same threshold as the VOI.