摘要

The copepod Acartia tonsa (Dana) is a promising live feed species for marine larviculture. A considerable progress has been made on the culturing technique for mass production, yet the application of automatic systems for quantification of A. tonsa development stages and biomass has largely been ignored. In this study, we propose a novel method for highly precise classification of development stages and biomass of A. tonsa, in intensive live feed cultures, using an automated ZooImage system, a freeware image analysis. We successfully created a training set of 13 categories, including 7 copepod and 6 non-copepod (debris) groups. ZooImage used this training set for automatic discrimination through a random forest algorithm with the general accuracy of 92.8%. The ZooImage showed no significant difference in classifying solitary eggs, or mixed nauplii stages and copepodites compared to personal microscope observation. Furthermore, ZooImage was also adapted for automatic estimation of A. tonsa biomass. This is the first study that has successfully applied ZooImage software which enables fast and reliable quantification of the development stages and the biomass of A. tonsa. As a result, relevant culturing parameters in intensive A. tonsa culture such as hatching success, mortality rate, development rate, and reproduction output of A. tonsa can be calculated automatically. Moreover, ZooImage is set up with an inexpensive desktop scanner and computer that can easily be applied in aquaculture hatcheries.

  • 出版日期2014-2-20