Quantification of ovarian follicles in bluefin tuna Thunnus thynnus by two stereological methods

Transcription

1 Journal of Fish Biology (2010) 77, doi: /j x, available online at Quantification of ovarian follicles in bluefin tuna Thunnus thynnus by two stereological methods L. Aragón, G. Aranda, A. Santos and A. Medina* Departamento de Biología, Facultad de Ciencias del Mar y Ambientales, Avda. República Saharaui, s/n, E-11510, Puerto Real, Cádiz, Spain (Received 3 August 2009, Accepted 13 May 2010) The numbers of different types of ovarian follicles (developing, degenerating and postovulatory follicles) were estimated in bluefin tuna Thunnus thynnus using two stereological procedures: the model-based method of Weibel & Gomez, which has become a tool of broad application in the quantification of oocytes in fishes, and the assumption-free disector (sic) method of Sterio. The estimates of developing follicles (follicles containing lipid-stage, vitellogenic and migratory-nucleus oocytes) made by the model-based method tended to be lower than those obtained with the disector, though significant differences were not observed except for vitellogenic follicles. Counts of atretic follicles by the model-based method were higher than those made using the disector, the differences being remarkable between both techniques, particularly in the case of β-atresia, where the statistical analysis indicated significantly unequal estimations with the two methods. In contrast, the amount of postovulatory follicles estimated by the disector, which would stand for the realized batch fecundity, was somewhat larger than that calculated with the model-based method. Journal compilation 2010 The Fisheries Society of the British Isles Key words: fecundity; histology; ovary; stereology. INTRODUCTION As with many fish species, bluefin tunas exhibit an asynchronous pattern of oocyte development and are multiple spawners that can shed several batches of eggs in the course of a single reproductive season (Schaefer, 2001). Two types of fecundity, determinate and indeterminate, have been defined in fishes on the basis of the strategy of recruitment of oocytes to the stock of mature oocytes (Murua & Saborido-Rey, 2003). While in fish species with determinate fecundity the potential annual fecundity can be estimated in late maturing females prior to the onset of spawning, in fishes with indeterminate fecundity, such as tunas (Scombridae), the annual fecundity should be calculated from the estimated number of oocytes spawned per batch and the number of spawnings carried out throughout the reproductive season (Murua & Saborido-Rey, 2003; Murua et al., 2003). Estimates of fecundity (potential fecundity) are often obtained from counts of the most advanced (large yolked) oocytes in the ovary, either by the gravimetric method, stereology *Author to whom correspondence should be addressed. Tel.: ; fax: ; antonio.medina@uca.es 719 Journal compilation 2010 The Fisheries Society of the British Isles

2 720 L. ARAGÓN ET AL. applied to histological sections, or automated particle counting. Data on the number of large yolked oocytes, however, may significantly overestimate the actual fecundity (realized fecundity), because a proportion of these oocytes can enter atresia and be reabsorbed before spawning (Kjesbu et al., 1991; Óskarsson et al., 2002; Kurita et al., 2003). A more accurate approach to determining the exact number of eggs released at a spawning event is to count the number of fresh postovulatory follicles. Stereology is a precise tool for obtaining quantitative information about threedimensional microscopic structures from planar sections. The first practical stereological technique was based on the Delesse principle (Delesse, 1847), which states that the fractional volume of a component in a tissue is proportional to its fractional cross-sectional area. Underlying this principle is the assumption that the target component is distributed randomly or evenly throughout the tissue. Using the method of Weibel & Gomez (1962), the stereological principle has been widely and successfully employed in estimating oocyte numbers in ovaries of various fish species (Emerson et al., 1990; Coward & Bromage, 2002; Medina et al., 2002, 2007; Murua et al., 2003; Cooper et al., 2005). This stereological approach, however, depends on assumptions about particle shape and size distribution and is consequently biased. Indeed a major drawback of this method arises when dealing with particles that are highly irregular in shape, such as atretic or postovulatory follicles (POF). Hence, degenerating follicles and POF are inherently difficult to quantify due to their irregular shape and a wide fluctuation in their size; these factors obviously limit the effectiveness of model-based techniques. On the contrary, the disector (sic) method (Sterio, 1984) is inherently unbiased (Howard & Reed, 2005) but is too labour costly if it is to be used routinely in studies involving large samples. It is thus of interest to check whether the model-based method that has been used traditionally in quantitative studies on fish reproduction provides results equivalent to those obtained with the assumption-free procedure. In this article, the number of different types of ovarian follicles is estimated in T. thynnus, using the two stereological methods. For evaluation of developing follicles, three distinct oocyte developmental stages were considered: lipid-stage, vitellogenic and migratory-nucleus oocytes. In addition, degenerating follicles in α and β stages of atresia, as well as freshly ovulated follicles (postovulatory follicles), were quantified by both stereological methods. The data obtained with the application of the method of Weibel & Gomez (1962) are compared to that produced by the physical disector (Sterio, 1984) in order to ascertain whether both procedures provide similar results for the different follicle types. MATERIALS AND METHODS SAMPLE COLLECTION Sexually mature female Atlantic T. thynnus were sampled from commercial catches either during spawning migration at the Strait of Gibraltar or at the spawning area in the western Sea (Table I). These samples include three different maturation phases of the species reproductive cycle. Immediately after measuring the curved fork length L FC,the ovaries were removed and their mass recorded to the nearest 10 g (M G ). L FC was transformed into straight fork length (L F ) according to the formula L F = L FC, and the total body mass (M B ) was estimated from the functions: M B = L F, for trap-caught specimens, and M B = L F, for longline-caught specimens, according to

3 QUANTIFICATION OF THUNNUS THYNNUS OVARIAN FOLLICLES 721 Table I. Summary of data for the sampled Thunnus thynnus. According to the histological structure of the ovary, the specimens were classified as active non-spawning (ANS), active spawning (AS) and inactive mature (IM) Fish ID Date of capture Location Fishing gear L F (cm) M G (g) Reproductive phase I G T1 5 June 2003 Strait of Gibraltar Trap ANS 1 41 T2 5 June 2003 Strait of Gibraltar Trap ANS 1 47 LL1 13 May 2006 Western Longline ANS 2 61 LL2 22 July 2006 Western Longline ANS 1 60 LL3 7 May 2006 Western Longline IM 1 55 LL4 17 June 2006 Western Longline ANS 8 24 LL5 9 May 2006 Western Longline ANS 0 77 PS1 29 June 2000 Western Purse seine AS 5 25 PS2 1 July 2000 Western Purse seine AS 4 41 L F, fork length; I G, gonado-somatic index; M G, ovary mass. conversion factors of the International Commission for the Conservation of Atlantic Tunas (ICCAT; The gonado-somatic index (I G ) was calculated as percentage of M G relative to M B, I G = 100 M G MB 1. The ovarian volume (V G) was estimated from M G according to the equation: V G = M G (Medina et al., 2007). HISTOLOGY For histological sampling, a slice of c. 0 5 cm thick was cut from the central part of one of the ovaries of each specimen. A portion of this slice, comprising the whole ovarian wall, was removed and fixed for at least 24 h in 4% formaldehyde (10% formalin) in phosphate buffer, 0 1 M, ph 7 2. Afterwards, the tissue samples were washed in buffer, dehydrated through increasing concentrations of ethyl alcohol, cleared in xylene and embedded in paraffin. Serial 10 μm sections were cut with a microtome Leica-Jung RM2025 (www. leica-microsystems.com), stained with heamatoxylin-vof (Gutiérrez, 1967) and mounted with Eukitt ( for microscopical observation. The histological sections were photographed on a light microscope Nikon Eclipse, 90i equipped with a digital photomicrographic system (Nikon Digital Camera Dxm 1200F; The micrographs were taken with either 4 (Weibel & Gomez method) or 2 (disector method) objective magnifications. A calibration scale was photographed at the same magnifications for calculation of the real dimensions of the structures in the corresponding images. The same series of histological sections were utilized for both stereological techniques. Four distinct types of developing follicles were distinguished depending on the respective oocyte maturation stages [Fig. 1(a)]: perinucleolar (PNF), lipid-stage (LSF), vitellogenic (VF) and migratory-nucleus (MNF) follicles. Furthermore, when present, another three types of follicles were assessed: α-stage (αaf) and β-stage (βaf) atretic follicles [Fig. 1(b)] and POF [Fig. 1(c)] (Medina et al., 2007). According to the extent of atresia and the most advanced group of oocytes present in the ovary, the individuals were classified into three reproductive phases (Table I): inactive mature, active non-spawning and active spawning (Schaefer, 1996, 2001).

4 722 L. ARAGÓN ET AL. (a) (b) (c) Fig. 1. Thunnus thynnus ovaries showing (a) follicles with diverse stages of oocyte development, (b) atretic follicles and (c) postovulatory follicles. LSF, lipid-stage follicles; MNF, migratory-nucleus follicles; POF, postovulatory follicles; VF, vitellogenic follicles; αaf, α-atretic follicles; βaf, β-atretic follicles. Scale bars: (a) 1 mm, (b) and (c) 500 μm. STEREOLOGY Model-based method Numbers of the different categories of follicles were calculated according to the formula of Weibel & Gomez (1962): N V = Kβ 1 NA 1 5V V 0 5 where N V is the numerical density (number per unit volume) of the considered follicle category, β is a shape coefficient, K is a size distribution coefficient, N A is the number of follicle transections per unit area and V V is the partial area (volume density). All the particles counted were considered ellipsoidal for calculation of β according to the graph provided by Weibel (1969). β and K were determined as previously indicated by Medina et al. (2002, 2007); values of these coefficients for each follicle type are shownintableii.v V was calculated by image analysis from 10 digital micrographs from each ovary through the software ImageJ ( (Medina et al., 2007). The total number of follicles contained in the ovaries (N) was obtained extrapolating N V to the

5 QUANTIFICATION OF THUNNUS THYNNUS OVARIAN FOLLICLES 723 Table II. Values of the coefficients K and β applied for the different follicle types Follicle developmental stage K β Lipid stage Vitellogenic stage Final maturation stage α-atresia β-atresia Postovulatory follicle total volume of the two ovaries: N = N V V G. Corrections of volume reduction due to the histological processing of samples were not applied in the present comparative study. Disector Two micrographs were taken from each histological section considered. Numbers of follicles containing oocytes at different developmental stages (LSF, VF and MNF), as well as α- andβ-atretic and postovulatory follicles (αaf, βaf and POF), were estimated according to the physical disector procedure. The disector principle (Sterio, 1984) requires two adjacent sections (disector pair) to be a known distance apart. The appropriate separation between sections (h) depends on the particle size; it should lie between 0 25 and 0 33 of its mean small diameter. For convenience, in all developing follicles, the elements chosen as the counting particles were the oocyte nuclei instead of the entire follicles, because the considerably smaller size of the nuclei allows for more accurate and faster alignment of the sections of the disector pair. In the particular cases of POF and atretic follicles, which lack a distinct oocyte nucleus, the whole elements were counted. In practice, the chosen separations between sections were 20 μm for nuclei of LSF, 30 μm for nuclei of VF and MNF, 100 μm for whole αaf and 40 μm for whole βaf and POF. A disector counting frame, equivalent to a true area of mm 2 was placed randomly onto each of the two images taken from the reference section (here conventionally referred to as section # 1 in the series of 10 μm consecutive sections), and then onto the look-up section [the section number of which depends on the type of follicle to be counted; e.g. #3forLSF(h = 20 μm), # 4 for VF (h = 30 μm) and so on]. Previously, both sections had been precisely aligned in order to make reliable comparisons within the working area between the two planes (Fig. 2). Ten additional frames were similarly established using consecutive sections obtained at five different depths in the paraffin block, making a total of 12 counting frames per ovary. Following the rules of this method, particles (either nuclei or whole follicles) that appeared in the counting frame on the reference section but not in the look-up section were counted. When the particles touched the forbidden lines of the counting frames, they were not counted (Sterio, 1984; Gundersen el al., 1988; Mayhew & Gundersen, 1996; Andersen, 2003; Howard & Reed, 2005). The numerical density of follicles (number per gonad unit volume) was estimated as follows: N V = Q a 1 P 1 h 1,whereh is the disector thickness and ranges between 20 and 100 μm, depending on the follicle type, a is the area of the rectangular reference frame of each disector (here, mm 2 ), P is the total number of disectors used for a gonad (here, 12 in most cases), Q is the total number of follicles counted in the P disectors in a gonad, namely the added number of oocyte nuclei or entire follicles present in the reference frames but not in the look-up sections. Finally, the estimate of the total oocyte number was computed as N = V G N V. It was estimated that the development of the disector method takes about three times longer than the model-based method. STATISTICAL ANALYSIS To test potential differences in the counts produced by the two stereological procedures in each individual, the paired-sample t-test, with α = 0 05, was applied to the data recorded for each follicle type in the nine fish examined (Zar, 1996).

6 724 L. ARAGÓN ET AL. (a) (b) Fig. 2. Illustration of one of the disectors used in this study. The counting frame area is a = mm 2 and the distance between the sections h = 30 μm (a). The nucleus of vitellogenic oocyte A is seen in the reference section, but (b) has disappeared in the look-up section; therefore, this particle was considered in the quantification of vitellogenic follicles. Follicle B, however, was not counted, since the oocyte nucleus is cut by one of the red (forbidden) lines of the frame. Scale bar: 1 mm. RESULTS Differences in the estimates between the two stereological methods [Weibel & Gomez (1962) minus disector] are shown in Table III, where they are expressed also as percentages (Z), according to the formula: Z = 100 (x y)x 1,wherex and y are the higher and lower value. The counts of developing follicles (LSF, VF and MNF) obtained with the model-based method of Weibel & Gomez (1962) tended to be lower than those calculated by the disector, with differences ranging between 2 1% (VF in LL4) and 31 4% (VF in LL3). An exception was LL2, in which the disector estimates of both LSF and VF were lower, whereas in individuals T1 and T2 the model-based

7 QUANTIFICATION OF THUNNUS THYNNUS OVARIAN FOLLICLES 725 Table III. Quantification of different types of follicles [lipid stage (LSF), vitellogenic (VF), migratory-nucleus (MNF), α-stage (αaf), β-stage (βaf) and postovulatory (POF)] in Thunnus thynnus. The numerical density (N V ) and total number of follicles (N) were calculated from the equation of Weibel & Gomez (1962) (W-G) and the disector method (Sterio, 1984). The differences between both measurements (per cent in parentheses) are given N V (number of follicles mm 3 ) ID Follicle types W-G Disector W-G Disector Difference (%) T1 LSF (3 7) VF (24 7) αaf (35 8) βaf (48 2) T2 LSF (21 0) VF (4 6) αaf (9 9) βaf LL1 LSF (20 2) VF (19 0) αaf (14 7) βaf (67 6) LL2 LSF (2 0) VF (6 9) αaf (47 2) βaf (15 9) LL3 LSF (30 0) VF (31 4) αaf βaf (54 9) LL4 LSF (23 2) VF (2 1) αaf (67 5) βaf (72 5) LL5 LSF (26 4) VF (19 9) αaf (31 7) N

8 726 L. ARAGÓN ET AL. N V (number of follicles mm 3 ) Table III. Continued ID Follicle types W-G Disector W-G Disector Difference (%) βaf PS1 LSF (9 3) VF (25 3) MNF (10 4) αaf βaf (15 8) POF (29 1) PS2 LSF (10 7) VF (19 3) MNF (10 4) αaf βaf (78 8) POF (10 8) N method overestimated the number of LSF compared to the disector (Table III). In general, the counts of LSF, VF and MNF obtained with both techniques were similar (Fig. 3), though the paired-sample t-test stated that the two procedures gave significantly unequal measures for VF. In individuals PS1 and PS2, the potential batch fecundity can be determined as the total number of MNF. Counts of αaf by the model-based method in specimens LL1, LL2, LL4 and LL5 were higher than those made using the disector, the differences being relatively high (67 5% in LL4) (Table III); in contrast, the disector estimated a greater number of αaf in T1 and T2. No significant statistical differences were found for this type of atretic follicle between both methods. Similarly, the model-based method overestimated the amount of βaf in relation to the disector technique in all cases but one (LL2) (Fig. 3), both estimates differing between 15 8% (PS1) and as much as 78 82% (PS2). The statistical analysis showed that such discrepancies were significant (P <0 05). When POF were present, the quantities estimated by the disector were larger than those calculated with the model-based method (Table III and Fig. 3), though the differences were not statistically significant (P >0 05). The estimated numbers of POF would stand for the realized batch fecundity of specimens PS1 and PS2. In these individuals, the disector produced counts of POF that were c. 20% (PS2) and 25% (PS1) lower than the estimated potential batch fecundity, as represented by the counts of MNF. DISCUSSION An accurate assessment of reproductive variables, such as age at maturity and annual fecundity, is crucial for the understanding of the population dynamics of fishes. In terms of reproductive strategy, T. thynnus, like other tunas, shows

9 QUANTIFICATION OF THUNNUS THYNNUS OVARIAN FOLLICLES (a) T1 T2 LL1 LL2 LL3 LL4 LL5 PS1 PS (d) αaf T1 T2 LL1 LL2 LL3 LL4 LL5 PS1 PS (b) VF Number of follicles ( 10 6 ) T1 T2 LL1 LL2 LL3 LL4 LL5 PS1 PS2 (e) T1 T2 LL1 LL2 LL3 LL4 LL5 PS1 PS (c) MNF T1 T2 LL1 LL2 LL3 LL4 LL5 PS1 PS (f) T1 T2 LL1 LL2 LL3 LL4 LL5 PS1 PS2 Fig. 3. Counts of the different follicle types: (a) lipid stage, (b) vitellogenic, (c) migratory-nucleus, (d) α-atretic, (e) β-atretic and (f) postovulatory obtained with the two stereological procedures, Weibel & Gomez (1962) ( ) and disector (Sterio, 1984) ( ). The highest differences are seen in atretic follicles.

10 728 L. ARAGÓN ET AL. asynchronous oocyte development, releasing several batches of eggs during the reproductive season and exhibiting indeterminate fecundity. In fish species that have indeterminate fecundity, the annual fecundity needs to be calculated from the combination of data about batch fecundity, spawning frequency (or interspawning interval) and duration of the spawning season (Murua et al., 2003). As spawning in T. thynnus occurs at sea water temperatures >24 C, the postovulatory follicles are assumed to be short-lived and reabsorbed within 24 h after spawning, such as has been determined for Katsuwonus pelamis (L.) (Hunter et al., 1986) and Thunnus albacares (Bonnaterre) (Schaefer, 1996). Based on the postovulatory follicle method (Hunter & Macewicz, 1985), the spawning frequency in actively spawning populations of T. thynnus has been estimated as c spawnings per day, which corresponds to an average interspawning interval of 1 2 days (Medina et al., 2002, 2007). These data, in conjunction with estimations of the individual duration of the spawning period by means of electronic tagging or observation on cage-reared broodstocks, may serve to estimate the number of egg batches shed over a spawning season. Stereometry is a powerful procedure for quantifying particles in three-dimensional samples. The stereological principle has been applied to ovarian histological sections of fishes as an accurate tool for fecundity estimates. In this context, the most widely used stereological method has been the one developed by Weibel & Gomez (1962) (Emerson et al., 1990; Coward & Bromage, 2002; Medina et al., 2002, 2007; Murua et al., 2003; Cooper et al., 2005). Nevertheless, it has the disadvantage of being subjected to assumptions about particle shape and size distribution. Design-based (unbiased) stereology requires a systematic uniform random sampling ensuring that any part of the studied organ has the same chance of being sampled (Howard & Reed, 2005). When uniform random samplings become impractical, unbiased estimations should rely on the assumption that the distribution of the quantified particles throughout the organ is homogeneous, so that the sub-samples to be examined represent the overall structure of the organ. In this study, a design-based sampling was not feasible because of strong restrictions met onboard fishing vessels; the method employed is, however, adequate for the present comparative analysis between stereological methods, since these were performed on the same samples. Previous estimations of fecundity in T. thynnus have been made from stereological counts of advanced vitellogenic oocytes (i.e. large yolked or migratory-nucleus oocytes), which define the potential fecundity. These data are thought to overestimate the actual (realized) fecundity because it does not consider follicular atresia. Therefore, a more straightforward way to determine the realized batch fecundity is to quantify the POF, which are the follicle remnants left after release of mature oocytes. As POF are irregular in shape, quantification of these structures from ovary histological samples should be performed preferably by assumption-free stereology. The volume loss caused by dehydration during the histological processing of the tissue has not been considered in this study, which obviously leads to overestimation of the numerical density (N V ). In similar samples, Medina et al. (2007) calculated a volume reduction of nearly 28% in the initial volume and applied a correction factor of 0 72 to the data obtained for N V and N. Counts of developing follicles (LSF, VF and MNF) provided by the model-based method of Weibel & Gomez (1962) were lower than those obtained by the disector, though significant differences were found only for VF. The total number of MNF in PS1 and PS2 would represent the potential batch fecundity.

11 QUANTIFICATION OF THUNNUS THYNNUS OVARIAN FOLLICLES 729 In contrast to the results obtained with developing follicles, numbers of atretic follicles were clearly overestimated by the model-based method with respect to the disector technique, especially in βaf, where the statistical analysis revealed significant differences between methods, probably because of the highly irregular shape and size of these particles, which depart considerably from regular ellipsoids making the model-based method inadvisable. The numbers of POF calculated by the modelbased method were lower than the disector estimates in the two individuals where these structures were present, but the differences between them were not significant. Further studies involving a larger sample of POF-containing ovaries are necessary to strengthen the statistical analysis. In conclusion, the assumption-based method tended to provide slightly lower counts than the disector in developing follicles, producing statistically similar results except for VF. The numbers of atretic follicles, in particular βaf, however, were clearly overestimated by the model-based method with respect to the disector. Although assumption-free stereology is recommended for unbiased quantitative estimations of irregularly shaped particles, the statistical analysis failed to detect significant differences for POF and αaf. Larger samples are therefore needed before definitive conclusions can be drawn. We greatly appreciate the extremely useful comments of three anonymous reviewers on an earlier version of this article. This work has been jointly funded by the Spanish Ministry of Science and Innovation (CTM C02-01/MAR), Junta de Andalucía (RNM-02469) and Red Eléctrica de España Fundación Migres. References Andersen, T. E. (2003). Unbiased stereological estimation of cell numbers and volume fractions: the disector [sic] and the principles of point counting. In Modern Approaches to Assess Maturity and Fecundity of Warm- and Cold-Water Fish and Squids (Kjesbu, O. S., Hunter, J. R. & Witthames, P. R., eds), pp Bergen: Institute of Marine Research. Cooper, D. V., Pearson, K. E. & Gunderson, D. R. (2005). Fecundity of shortspine thornyhead (Sebastolobus alascanus) and longspine thornyhead (S. altivelis) (Scorpaenidae) from the northeastern Pacific Ocean, determined by stereological and gravimetric techniques. Fishery Bulletin 103, Coward, K. & Bromage, N. R. (2002). Stereological point-counting; an accurate method for assessing ovarian function in tilapia. Aquaculture 212, Delesse, M. A. (1847). Procédé mécanique pour déterminer la composition des roches. Comptes Rendus de l Académie des Sciences de Paris 25, Emerson, L. S., Greer Walker, M. & Witthames, P. R. (1990). A stereological method for estimating fish fecundity. Journal of Fish Biology 36, Gundersen, H. J. G., Bagger, P., Bendtsen, T. F., Evans, S. M., Korbo, L., Marcussen, N., Møller, A., Nielsen, K., Nyengaard, J. R., Pakkenberg, B., Sørensen, F. B., Vesterby, A. & West. M. J. (1988). The new stereological tools: disector [sic], fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis. APMIS 96, Gutiérrez, M. (1967). Coloración histológica para ovarios de peces, crustáceos y moluscos. Investigación Pesquera 31, Howard, C. V. & Reed, M. G. (2005). Unbiased Stereology: Three-Dimensional Measurement in Microscopy. Abingdon: Garland Science/BIOS Scientific Publishers. Hunter, J. R. & Macewicz, B. J. (1985). Measurement of spawning frequency in multiple spawning fishes. In An Egg Production Method for Estimating Spawning Biomass of Pelagic Fish: Application to the Northern Anchovy, Engraulis mordax (Lasker, R., ed.), pp NOAA Technical Report NMFS 36.

12 730 L. ARAGÓN ET AL. Hunter, J. R., Macewicz, B. J. & Sibert, J. R. (1986). The spawning frequency of skipjack tuna, Katsuwonus pelamis, from the South Pacific. Fishery Bulletin 84, Kjesbu, O. S., Klungsøyr, J., Kryvi, H., Witthames, P. R. & Greer Walker, M. (1991). Fecundity, atresia, and egg size of captive Atlantic cod (Gadus morhua) in relation to proximate body composition. Canadian Journal of Fisheries and Aquatic Sciences 48, Kurita, Y., Meier, S. & Kjesbu, O. S. (2003). Oocyte growth and fecundity regulation by atresia of Atlantic herring (Clupea harengus) in relation to body condition throughout the maturation cycle. Journal of Sea Research 49, Mayhew, T. M. & Gundersen, H. J. G. (1996). If you assume, you can make an ass out of u[sic] and me : a decade of the disector [sic] for stereological counting of particles in 3D space. Journal of Anatomy 188, Medina, A., Abascal, F. J., Megina, C. & García, A. (2002). Stereological assessment of the reproductive status of female Atlantic northern bluefin tuna, Thunnus thynnus (L.), during migration to spawning grounds through the Strait of Gibraltar. Journal of Fish Biology 60, doi: /jfbi Medina, A., Abascal, F. J., Aragón, L., Mourente, G., Aranda, G., Galaz, J. M., Belmonte, A., de la Serna, J. M. & García, S. (2007). Influence of sampling gear in assessment of reproductive parameters for bluefin tuna in the western. Marine Ecology Progress Series 337, Murua, H. & Saborido-Rey, F. (2003). Female reproductive strategies of marine fish species of the North Atlantic. Journal of Northwest Atlantic Fisheries Science 33, Murua, H., Kraus, G., Saborido-Rey, F., Witthames, P. R., Thorsen, A. & Junquera, S. (2003). Procedures to estimate fecundity of marine fish species in relation to their reproductive strategy. Journal of Northwest Atlantic Fisheries Science 33, Óskarsson, G. J., Kjesbu, O. S. & Slotte, A. (2002). Predictions of realized fecundity and spawning time in Norwegian spring-spawning herring (Clupea harengus). Journal of Sea Research 48, Schaefer, K. M. (1996). Spawning time, frequency, and batch fecundity of yellowfin tuna, Thunnus albacares, near Clipperton Atoll in the eastern Pacific Ocean. Fishery Bulletin 94, Schaefer, K. M. (2001). Reproductive biology of tunas. In Tuna: Physiology, Ecology and Evolution (Block, B. A. & Stevens, E. D., eds), pp London: Academic Press. Sterio, D. C. (1984). The unbiased estimation of number and sizes of arbitrary particles using the disector [sic]. Journal of Microscopy 134, Weibel, E. R. (1969). Stereological principles for morphometry in electron microscopy cytology. International Review of Cytology 26, Weibel, E. R. & Gomez, D. M. (1962). A principle for counting tissue structures on random sections. Journal of Applied Physiology 17, Zar, J. H. (1996). Biostatistical Analysis. Upper Saddle River, NJ: Prentice-Hall.