An automated image analysis system for estimating fecundity in portunid crabs

Abstract

Robust assessment of exploited crustacean stocks requires estimates of reproductive output in relation to size and environmental variability. An automated, high-throughput image analysis approach was developed for estimating fecundity in crabs (using Blue Swimmer Crab, Portunus armatus). The approach employed a waterproofed flatbed scanner, whereby images of egg samples were digitised, followed by automated analysis of each image using customised computer macros. Paired counts (of the same samples) were also conducted using traditional counting of eggs under a compound microscope. Trials demonstrated that the high-throughput technique was 6 times faster than the microscope technique, and simultaneously provided information on egg size. Egg count data collected using the high-throughput technique was modelled to inform the collection of samples for estimation of fecundity in Blue Swimmer Crab. This simulation found that conducting egg counts on as few as 10 egg masses per month produced a 90% probability that the estimate of mean eggs-per-gram-of-egg-mass was within 15% of the actual mean (for upscaling across the population). Quantifying temporal variation in fecundity, and potential environmental drivers of this variation, is important in the assessment of crab stocks and ensuring sustainable commercial and recreational harvest. This method of counting eggs has broad applications for other highly fecund, egg-bearing species.