Stereological calibration of the profile method to quickly estimate atresia levels in fish

The (physical) ‘disector method’, a frequently cited stereological technique, has so far received little attention within fish fecundity studies, although it can be used to provide unbiased, assumption-free data on levels of atresia (resorption of vitellogenic oocytes). In comparison, traditional simple counting to estimate the numerical ratio of normal to atretic cells is biased because the smaller atretic cells have a lower chance of being transected in histological sections. These problems are circumvented by the disector method as it operates in three dimensions, i.e., uses consecutive pairs of slides. However, the high labour costs involved prevent regular usage of this method in population studies where large numbers of ovarian samples are processed. In the present paper we assess the corresponding bias of the traditional profile method, analysing developing ovaries of Atlantic cod (Gadus morhua) and herring (Clupea harengus), i.e., in a relatively broad and narrow oocyte frequency situation, respectively. A highly significant but non-linear relationship (r2 = 0.975, P < 0.001, df = 154) was found between the relative intensity of atretic vitellogenic oocytes (ARI) estimated by the disector and profile method. Both species fitted well to this polynomial model. The degree of underestimation of atretic oocytes still containing yolk (i.e., the alpha stage) was at maximum (11.6%) at intermediate levels (ARI = 50% (disector)) but, logically, no such bias existed at the extreme ends (ARI: 0% and 100% (disector)). The practical use of this simple, fast approach designated the stereo-profile method, including the additional use of image analysis for further refinements, is successfully demonstrated on field samples.