Fecundity estimation by oocyte packing density formulae in determinate and indeterminate spawners: Theoretical considerations and applications

Abstract

This paper explores why the ‘Auto-diametric method’, currently used in many laboratories to quickly estimate fish fecundity, works well on marine species with a determinate reproductive style but much less so on species with an indeterminate reproductive style. Algorithms describing links between potentially important explanatory variables to estimate fecundity were first established, and these were followed by practical observations in order to validate the method under two extreme situations: 1) straightforward fecundity estimation in a determinate, single-batch spawner: Atlantic herring (AH) Clupea harengus and 2) challenging fecundity estimation in an indeterminate, multiple-batch spawner: Japanese flounder (JF) Paralichthys olivaceus. The Auto-diametric method relies on the successful prediction of the number of vitellogenic oocytes (VTO) per gram ovary (oocyte packing density; OPD) from the mean VTO diameter. Theoretically, OPD could be reproduced by the following four variables; OD V (volume-based mean VTO diameter, which deviates from arithmetic mean VTO diameter), VFvto (volume fraction of VTO in the ovary), ρ o (specific gravity of the ovary) and k (VTO shape, i.e. ratio of long and short oocyte axes). VF vto, ρ o and k were tested in relation to growth in OD V. The dynamic range throughout maturation was clearly highest in VF vto. As a result, OPD was mainly influenced by OD V and secondly by VFvto. Log (OPD) for AH decreased as log (OD V) increased, while log (OPD) for JF first increased during early vitellogenesis, then decreased during late vitellogenesis and spawning as log (OD V) increased. These linear regressions thus behaved statistically differently between species, and associated residuals fluctuated more for JF than for AH. We conclude that the OPD-OD V relationship may be better expressed by several curves that cover different parts of the maturation cycle rather than by one curve that cover all these parts. This seems to be particularly true for indeterminate spawners. A correction factor for vitellogenic atresia was included based on the level of atresia and the size of atretic oocytes in relation to normal oocytes finding that OPD would be biased when smaller atretic oocytes are present but not accounted for. Furthermore, special care should be taken when collecting sub-samples to make them as representative as possible of the whole ovary, including in terms of relative amount of ovarian wall and stroma. Theoretical consideration, along with original, high-quality information regarding the above-listed variables made it possible to reproduce very accurately the observed changes in OPD, but not yet precisely enough at the individual level in indeterminate spawners.