This is something I have been wondering about for a long time. How does the way in which we count copepods (and to a lesser extent, larvaceans) affect when an image is considered "complete" in terms of its identification? I've also wondered how our methods affect the biomass estimate?
Different classifiers have different comfort levels with identifying the smaller/blurrier organisms. I know for myself, I have tried to hold myself to the standard of only marking copepods when I see the antennae visible. However, I end up breaking my own rule all the time, because sometimes the antennae are not visible (when a copepod is in the folded-over, shrimp-like "escape" posture, for example, or when it is in the "jumping position".)
I personally use many features to help me decide whether to mark small or blurry organisms. Some things I consider are the overall shape of the organism; whether it is the CAL or MED data set; how much marine snow is in the frame; and if there are other copepods/larvaceans/houses present in the frame.
Since there are relatively few active classifiers on the Plankton Portal project and it takes five agreements on identifications to mark a frame as "complete," it might be good to develop some consistency around this. In other words, in order to get five classifiers to agree on the exact number of copepods in a frame might take a very long time. Or is the frame marked "complete" when five classifiers mark a minimum number of copepods, even though some other classifiers might have marked above that number? Say five classifiers mark three copepods, while three others mark 5, 7, or 8 copepods. Does the frame go down in the data as containing three copepods?
Copepods are clearly underestimated in the biomass estimate. For many of the copepods we mark, we see many others that are "probably but not definitely" copepods. How do you account for this in the data? Or do you assume that the biomass of all of the organisms are underestimated to the same degree due to the limitations of sampling by imaging?