Research Note: A validation of an image-based method to estimate chicken comb size

Abstract

Chicken combs carry important information about the individual, especially the size has been related to sexual reproduction, health, and social signaling. Comb size is usually estimated by weighing removed combs or by calculating the product of the comb's longest and highest dimensions (LHA) to approximate comb area based on measures of a ruler or caliper. These methods have several shortcomings including invasiveness or imprecision. As a result, more recent efforts have employed pixel-based approximations of comb area (PBA) from images. However, the validity of PBA to estimate comb area and how the approximation compares to previous approximation methods, such as LHA, is unknown. Therefore, we developed an apparatus for taking standardized images of the head position of the hens and then applied PBA using the software ImageJ. The hens were each photographed 3 times by 2 different handlers. We first tested the accuracy of the pixel-based area approximation on 3 geometric shapes of known area. Second, we tested the precision of PBA of 15 hens (Dekalb White), evaluated as within-image and within-individual hen precision. Furthermore, we compared the PBA with the LHA based on measures of a caliper. The PBA was both accurate and precise, whereas the LHA overestimated comb area with increasing overestimation for larger combs. Due to the greater accuracy of the PBA, as well as future possibilities of automation and inclusion of further measures, we suggest PBA as a more reliable approach to estimate comb area than LHA. Additionally, our results demonstrate that the outcomes of LHA should be evaluated on an ordinal scale level only.