Calibration of the repeatability of foraminiferal test size and shape measures with recommendations for future use

The fossil record of planktonic foraminifera is ideally suited to defining stratigraphic age controls and exploring fundamental questions in evolutionary biology due to its excellent preservation potential that yields continuous, high-resolution fossil archives of large numbers of individuals. For full morphometric analyses foraminifera tests are generally mounted, oriented and imaged manually, while data are processed using standard software such as ImageJ or Image Pro. However, manually induced orientation errors are a source of potential bias in trait measurements even when quantified using the same computational subroutine. Here we test the repeatability of four measures of foraminiferal test shape on six morphologically distinct species and present a calibration (power analysis) of the number of individuals needed to determine a given percentage change in these traits. We mounted and measured every individual twice and analysed the difference between the two measurements to determine the effects of small orientation changes on the studied traits. We show that measurements of test area and aspect ratio are statistically indistinguishable between runs for all species studied, and a power law calibration suggests that between 25 and 50 individuals are needed to detect at least a 10% in- or decrease in either trait. However, despite mounting tests on glass slides to clarify perimeter outlines, test perimeter was only repeatable in the spherical species Orbulina universa, and test roundness was not repeatable for three out of six studied species. We recommend the use of aspect ratios constructed from lengths and avoidance of perimeters and their dependent metrics to reduce orientation induced bias.