Witthames et al. : Advances in methods for determining fecundity in marine fishes 
151 
Table 2 
Details of the image analysis software PPilkington Image Analysis Systems, 2 Shareware available at rsb.info.nih.gov/ij/down- 
load.html), operating systems, camera, light source, microscope, resolution, grey scale, and staining methods used by each insti- 
tute: AZTI, Cefas, CSIC, and IMR to analyze whole mounts prepared from each species: Atlantic cod ( Gadus morhua ), European 
hake (Merluccius merluccius , Atlantic herring ( Clupea harengus ), Atlantic mackerel ( Scomber scombrus), European plaice ( Pleu - 
ronectes platessa), redfish (deep water redfish (Sebastes mentella ) or golden redfish ( Sebastes marinus). 
Detail 
AZTI [A] 
Cefas [B] 
CSIC [C] 
IMR [D] 
Software 
Visilog 6.11 
Myrmica 1 automatic or 
semi automatic operation 
QWin 
(Leica Imaging 
System) 
Image SXM v. 1.77 2 
Operating system 
PC Windows 
PC Windows 
PC Windows 
Mac OS X 
Camera 
Camedia-4040 Zoom 
Pulnix TMC-1000-CL 
Leica IC A 
JVC TK-1070E 
Light source 
3100 High light 
Olympus 
3100 Olympus High light 
Leica MZ6 
SZX-ILLB200 
Microscope 
Olympus SZX12 
Olympus SZX12 
Leica MZ6 
Olympus SZX12 
Resolution 
(pm/pixel) 
5 (hake) 
19.3 (plaice), 10.0 (cod), 
4.6 (mackerel) 
21.9-27.6 (redfish) 
7.02-22.43 (hake) 
14.7 (cod) 
14.7 (herring) 
Gray scale 
(255 saturated) 
230 
175 
90 
205 
Staining method 
Rose bengal (hake) 
PAS (mackerel and cod) 
Eosin (plaice) 
Rose bengal (hake) 
Unstained redfish 
Unstained 
analysis system and the size of the follicles comprising 
the fecundity. Institutes A and C split the sample and 
took a separate image of each aliquot to ensure that 
the follicles were evenly spread without overlap in a 
container that was completely covered within the field 
of view. Institute B spread the sample in a counting 
chamber 70 mm long and either 4, 7, or 10 mm wide for 
Atlantic mackerel, Atlantic cod, and European plaice, 
respectively. The three widths of counting chamber were 
used so that three magnifications (Table 2) could be 
used while still displaying the full width of the cham- 
ber. Above the sample the counting chamber tapered 
outwards in v-shaped profile leading to an upper liq- 
uid surface of 25 mm creating a flat meniscus over 
the channel holding the follicles. Myrmica software 
(Pilkington image analysis systems, Lindfield, West 
Sussex, UK) was used to create and archive a series 
of images and overlays along the horizontal axis of the 
sample container showing all measurements overlaying 
the follicles measured. Individual follicles stained PAS 
were measured in this counting chamber both manually 
and by image analysis to establish the accuracy of size 
measurement by image analysis. Institute D working 
with Atlantic cod and Atlantic herring used a method 
described previously (Thorsen and Kjesbu, 2001). 
In order to investigate whether fixing the ovary in 
sample aliquots or whole in the tunica affected mean 
follicular diameter (D f , pm), circularity, and fecundity 
per gram of ovary (F ow ), replicate samples were taken by 
pipette and scalpel, respectively, from the central part of 
the same ovary from Atlantic cod, Atlantic haddock, and 
European plaice (Table 1). These samples were collected 
from fish caught in the Irish Sea during February 2007 
(Table 1) and fixed for between 63 and 91 days in 1.7 
to 9 times their volume of NBF before image analysis 
(Table 2). Circularity, a function of follicular shape, 
was measured according to the following equation: 
Circularity = 4ji (area / perimeter) 2 . (1) 
Homogeneity of D f and F ow were studied in Atlantic 
cod and European hake ovaries to investigate whether 
a sample from the center of the ovarian mass was sig- 
nificantly different from samples taken at the extreme 
ends of each ovary or between the pair of ovaries. 
Stain evaluation 
Ovarian tissue (whole mounts) stained by the three 
routines (Table 2) was compared with nonstained tissue 
in order to improve the identification and measure- 
ment of developing (cortical alveoli and vitellogenic) 
and regressing (postovulatory and atretic) follicles. Non- 
stained tissue was prepared for analysis as described 
by Thorsen and Kjesbu (2001), and two of the staining 
methods applied water soluble 1% eosin or 0.02% rose 
bengal weight to volume (w/v) dissolved in NBF to color 
the follicles. A third staining method used the PAS 
reaction, previously applied to stain cortical alveoli fol- 
licles (Greer-Walker et al., 1994). In this procedure the 
concentration of PAS reagent was 0.1% and 15% w/v, 
respectively, compared to the histological procedure 
in order to minimize shrinkage of follicles. Nonbound 
stain was removed from the tissue subsamples after 
