Skip to main content
Dryad

Measuring avian bill size: Comparing and evaluating 3D surface scanning with traditional size estimates in Australian birds

Cite this dataset

Ryding, Sara et al. (2024). Measuring avian bill size: Comparing and evaluating 3D surface scanning with traditional size estimates in Australian birds [Dataset]. Dryad. https://doi.org/10.5061/dryad.sxksn038q

Abstract

Unidimensional measurements for estimating bill size, like length and width, are commonly used in ecology and evolution, but can be criticised due to issues with repeatability and accuracy. Furthermore, formula-based estimates of bill surface area tend to assume uniform bill shapes across species, which is rarely the case. 3D surface scanning can potentially help overcome some such issues by collecting detailed external morphology and direct measurements of surface area, rather than composite estimates of size. Here, we evaluate the use of 3D surface scanners on avian museum specimens to test the repeatability of 3D-based measurements and compare these to traditional formula-based methods of estimating bill size from unidimensional measurements. Using 28 Australian bird species, we investigate inter-observer repeatability of surface area measurements from 3D surface scans. We then compare 3D-based size estimates to formula-based size estimates to infer the accuracy and precision of formula-based measurements of bill surface area. We find that morphometric measurements from 3D surface scans are highly repeatable between observers, without the need for extensive training, demonstrating an advantage over unidimensional measuring methods, like callipers. When comparing 3D-based measurements to formula-based estimates of bill surface area, most formulae for estimating size consistently underestimate surface area, and with considerable variation between species. Where 3D scanning is not possible, we find that a commonly used cone formula for estimating bill size is most precise across diverse bill shapes, therefore supporting its use in interspecific contexts. However, we find that incorporating an additional unidimensional measure of bill curvature into formulae improves the accuracy of the calculated area. Our results reveal the high potential for 3D surface scanners in avian morphometric research, especially for studies necessitating large sample sizes collected by multiple observers, and gives suggestions for formula-based approaches to estimate bill size.

README: Measuring avian bill size: comparing and evaluating 3D surface scanning with traditional size estimates

https://doi.org/10.5061/dryad.sxksn038q

We investigated measurement repeatability between observers for bill surface area, as derived from 3D scans. We applied 3D surface scans on a diverse set of birds, and further collected unidimensional measurements. We assess how these unidimensional measurements can be used in different formulae to estimate bill surface area.

Description of the data and file structure

The dataset contains bill surface area measurements, extracted from two observers using 3D surface scans of diverse bird species (columns billSA_measurer1 and 2). It also contains unidimensional measurements (bill length, width, depth, width between the nares, and upper and lower bill profile length; as titled in columns), measured using calipers (or, in the case of total bill width and the profile length, derived from the 3D scans). All measurements are in mm. Finally, it contains important collection information: scientific name, common name, the catalogue number of the individual specimen, any comments on the specimen, its sex (male, female, or unknown), collection latitude and longitude, and collection year and month.

The data also contains the R code used to analyse the data (see below in the code section). 

Sharing/Access information

Data was derived from avian specimens from the following institutions:

  • Australian Wildlife Collection
  • Melbourne Museum
  • South Australian Museum
  • Australian Museum

Code/Software

The code in this submission analyses inter-observer measurement repeatability, a phylogenetic PCA, and calculations of AIC based off of various formula options. More details on statistical methods can be found in the publication, and the code is annotated. For any queries, contact Sara Ryding s.ryding@deakin.edu.au.

Code runs in R.

Funding

Australian Research Council, Award: DP190101244

Natural Sciences and Engineering Research Council, Award: RGPIN-2020-05089

Deakin University, Centre for Integrative Ecology