Variability in oocyte size and batch fecundity in anchoveta (Engraulis ringens, Jenyns 1842) from two spawning areas off the Chilean coast

1 laboratorio de oceanografía Pesquera y Ecología larval (loPEl), departamento de oceanografía, universidad de Concepción, P.o. box 160-C, Concepción, Chile. 2 Current address: instituto de Fomento Pesquero (iFoP), Valparaíso, Chile. E-mail: elson.leal@ifop.cl 3 Centro Fondap CoPas, universidad de Concepción, P.o. box 160-C, Concepción, Chile. 4 departamento de Ciencias del Mar, universidad arturo Prat, iquique, Chile.

1973 ;Funamoto and aoki, 2002;llanos-rivera and Castro, 2004).Hypotheses proposed to explain this variability include environmental factors such as temperature, oxygen and food availability (tanasichuk and Ware, 1987;beacham and Murray, 1993), and biological factors such as size and endocrine state of the female during oocyte growth in the ovary (Hay and brett, 1988;ojanguren et al., 1996;laine and rajasilta, 1999).because the initial egg size determines many features of the offspring and has an important effect on early life stage growth and survival (bagenal, 1971;Wootton, 1994;ojanguren et al., 1996;brooks et al., 1997), some of these hypotheses propose that egg size constitutes an adaptation to local environmental conditions.
the anchoveta Engraulis ringens is a partial spawner in the Humboldt Current with a marked winter (July-september) reproductive season along its wide geographic range (4º-42ºs).the species distribution and parasitic fauna (Valdivia et al., 2007) indicate the existence of two main spawning areas off Chile: one to the north (18º-23ºs) and one to the south (35-38ºs).separated by over 15 degrees of latitude, these spawning habitats show strong environmental differences.Castro et al. (2002) and llanos-rivera and Castro (2004Castro ( , 2006) ) reported a latitudinal gradient in egg size and other early life history traits of E. ringens along the Chilean coast.unfortunately, due to the planktonic origin of the eggs analysed, it was not possible to determine any relationship between female size (length and weight) and egg size in the plankton.also, it is not known whether these differences in egg size were the result of processes occurring inside the ovary (i.e. during the early oocyte stages, or later during hydration) or later when the eggs were released into the environment.
in this study, we consider whether intra-species variability in E. ringens egg number and oocyte size might be adaptive traits in individuals from the two spawning areas, where important differences in habitat conditions are observed.

MatErial and MEtHods
data for the analyses come from studies in which female size (length and weight), oocyte size, and sea surface temperature were measured simultaneously for a large set of female individuals.oocyte size (volume, mm 3 ) was measured from histological preparations of E. ringens ovaries obtained during fisheries assessment surveys in which the daily egg production method (dEPM) was used to estimate anchovy spawning biomass off northern (18-23°s and 70-72°W) and southern Chile (35-38°s and 72-73°W) (Fig. 1) in 2002, 2003 and 2004.the samples were collected during the peak spawning period (august-september) of each year.the females' (n>3000 in each year and area) total length (tl), total weight (tw) and ovary-free weight (ofw) were measured in both areas.
based on an ovarian development scale modified for E. ringens (Hunter and Macewicz, 1985), ovaries in the mature stage (with full yolk-stage oocytes) were selected under an optical microscope.since the techniques used for histological preparations deteriorate and deform the oocytes, only histological preparations that presented oocytes appropriate for measurement (oval form, oocyte and follicular cover integrity, central position and integrity of the nucleus) were selected.

Measurements and estimates
the selected histological preparations were digitally photographed at a screen resolution of 624 x 480 pixels with a sony CCd-iris video camera mounted on a microscope at 40x magnification.based on previous calibrations (120 pixels/1 mm), measurements done with digital oocyte images used image J software.the width and length of the oocyte and the nucleus were measured.oocyte volume (ov) was estimated assuming an ellipsoid shape of the E. ringens oocytes (llanosrivera and Castro, 2004), V = π * MaA*MiA2/6 (Maruyama et al., 2003), where Maa and Mia are the major and minor axes, respectively.to ensure that the analysed oocytes included only those cut on a longitudinal and central plane by the microtom, we only used oocytes whose nucleus size was within the standard deviation range estimated for each year and area.oocytes were at the same stage in both populations (full yolk-stage oocytes).based on these criteria, the number of oocytes measured per female ranged between 3 and 5.
batch fecundity (bf) (number of oocytes liberated by a female in one spawning event) was estimated for each female from a linear fecundity model based on ovary-free female weight (ofw) (Hunter and Macewicz, 1985) as PF (w) = a + b*Ofw; the parameters (a,b) were previously estimated for each year and area in the dEPM studies.the relationship between mean oocyte volume and batch fecundity was then estimated in each area.
the potential effect of female size (tl, tw) on oocyte volume was evaluated in females from each year and area.the possible differences in the female size among areas that could explain the variability in oocyte size was also analysed.Moreover, data on female length and weight in both areas were transformed to their natural logarithms, as proposed by Jonsson and Jonsson (1999), prior to comparing the regressions.
Previous reports indicated that the environmental temperature during the 90 to 60 days before spawning might have an important effect on fish fecundity and egg size (tanasichuk and Ware, 1987).accordingly, in this study we assessed the effect of the mean sea surface temperature 2 and 3 months (May, June) before the main spawning month (august) on the annual mean oocyte volume.the temperature value corresponded to the average between May and June and the standard deviation was the monthly variability each year.the mean monthly sea surface temperature in the coastal zone in 2002, 2003 and 2004 were taken from the Hydrographic and oceanographic service of the Chilean navy (sHoa) database.

Statistical analyses
Previous to all parametric analyses, the data were first tested for normal distribution (Kolmogorovsmirnov one-sample test (P>0.05)and homoscedasticity of variance (levene test, P>0.05).
one-way analysis of variance (anoVa) was used to evaluate differences in oocyte volume among years within each population (Power et al., 2005) and t-tests to determine whether differences in the mean oocyte volume occurred each year between areas.the t-tests were also used to assess whether differences in mean fecundity occur between stocks based on the three years of data.
the relationship between oocyte volume and batch fecundity was evaluated within each area through a Pearson's correlation coefficient.this analysis was also used to estimate the correlation between the females' size (tl, tw) and the oocyte volume for all the females measured in each area.to test the hypotheses about equality in regression coefficients of length-weight data of two populations, a t-test was used (zar, 1984).
a simple regression analysis was used to determine the correlation between mean sea surface temperature 2 and 3 months before the main spawning month on the annual mean oocyte volume within each area.
Finally, we carried out a two-way anoVa with interaction using area and year as factors to examine their effects on oocyte size in the areas included in the present study.rEsults the mean length and weight of the females sampled differed between populations (t-test, P<0.001), probably because too few small individuals (<13 cm tl) were included in the northern area (Fig. 2). the mean length and weight values were higher in the northern Chile population.the differences in the average tw persisted (t-test; P<0.001) when females were compared within the same length range (i.e.13-17.5 cm,) between populations.additionally, the regression analysis (Fig. 3) showed that northern females exhibited a higher mean weight than the southern ones for fish of the same length (t-test, P<0.001).
statistical analyses showed significantly different oocyte volumes (mm 3 ) between the northern and southern E. ringens populations (t-test, P<0.001).While in the northern area the oocyte sizes ranged between 0.005 and 0.045 mm 3 , in the south they ranged between 0.010 and 0.055 mm 3 (Fig. 4). in 2002, the mean oocyte volume was 95% larger in south Chile females (table 1); this difference decreased to 66% in 2003 and 30% in 2004 (Fig. 5).significant dif-ferences in mean oocyte volume were also detected among years within each area (anoVa, P<0.001) (table 2).
batch fecundity was also significantly different between areas.the mean batch fecundity (estimated from the ovary-free female weight) was higher (ttest; P<0.001) during the entire period for the northern females (12788 oocytes; std = 2175) than for the southern females (8197 oocytes, std = 2326).the differences in mean batch fecundity persisted among areas (t-test; P<0.001) when females were compared within the same length range (i.e.13-17.5 cm,) between populations.
the trends between ovary weight and ovary-free female weight were similar between females (Fig.      6a) from both areas.Pearson's correlation coefficient showed that oocyte volume was not significantly correlated (P>0.13) with fecundity in each area (Fig. 6b).also, oocyte volume was not significantly correlated with female length or weight (P>0.08).nevertheless, a trade off between oocyte volume and fecundity may be observed in the comparison of areas.sea surface temperature was the only environmental variable compared between the two spawning areas.the mean sea surface temperatures in May and June (two and three months prior to august, the peak spawning month) were lower (average 4°C every year) in the southern area than in the northern area and were not correlated with the mean oocyte volume (P>0.05)within each location (Fig. 7). the anoVa showed that area was the most important factor to explain the oocyte size variability (46%).the year effect and area x year interactions explained less variance (3.7 y 5.6%, respectively) but were also significant predictors of oocyte size variability in the geographical range and years included in this study (table 3).disCussion the general objective of this study was to determine whether differences in reproductive parameters and oocyte sizes occurred between E. ringens populations located along the Chilean coast and whether such intra-specific differences might be adaptive given the strong differences in spawning habitat conditions.the results of the present study show smaller oocyte volume and higher fecundity for the northern female anchoveta than for the southern ones.these differences were significant and persisted throughout the three-year study period.the results of this study agree with those obtained by llanos-rivera and Castro (2004), who analysed eggs obtained from plankton samples, reporting an increase in E. ringens egg volume with    latitude; the volumes were 55% smaller off iquique (northern Chile) than off talcahuano (southern Chile).the differences in oocyte size between populations found in the present study were 95% (2002), 66% (2003), and 30% (2004), and were thus in the middle of the range reported by llanosrivera and Castro (2004) for planktonic eggs.Variability in egg size has been described by several authors for diverse fish species that are distributed over wide geographic ranges or are exposed to divergent environmental conditions (de Ciechomski, 1973;tanasichuk and Ware, 1987;beacham and Murray, 1993;duponchelle et al., 2000;Funamoto and aoki, 2002;Kokita, 2003).batch fecundity, estimated from all female sizes sampled, was also significantly different between areas, with higher values for the northern females than for the southern ones.among females within the same body length range between populations, the differences in mean batch fecundity persisted among areas, indicating that the observed differences in bf cannot be attributed to differences in body sizes between populations.interestingly, when we compared our results with others from northern Peru (years 1981, 1985, 1990, 1994, 1995 and 1996) (ayon, 2000) a latitudinal trend seemed to emerge as batch fecundities in northern Peru seem to be higher (12701-18495 oocytes) than those we found in northern Chile (12788 oocytes, std = 2175) and much higher than those from southern Chile (8197 oocytes, std = 2326).
Egg size is affected by female body size in some fish (Chambers, 1997).However, in this case the oocyte volume was not correlated with female size within each population.these results, according to Kokita (2003), indicate that the effect of female size difference between populations can be ignored as a covariate to explain oocyte size variability between the two areas.in the literature, the relationship between female size and oocyte size in other small pelagic fish is not clear cut.some authors report this relationship for some species but not for others.For example, Hay and brett (1988) found a positive correlation between female length and/or age and egg weight in Clupea arengus, but Claramunt et al. (1994) reported that oocyte size in Sardinops sagax is independent of female body size.West (1990) indicated that, for small fish with a short life history, the narrow size range of the adults and their eggs could explain the lack of relationship between female size and egg size. in our study, oocyte volume was larger in southern females.since the ovary weight was similar between females from the two areas, the cost for spawning larger eggs in the southern females is a decrease in batch fecundity.a trade-off between fecundity and egg size has been reported before for other Engraulis species (Funamoto and aoki, 2002), suggesting that this trend may be a common feature but not extensively reported.in our study, though, we were unable to detect this inverse relationship within each population, and this could have been because bf were estimated from randomly selected females within the population and not based on direct measurements from the same females whose ov were measured.However, when we consider both populations combined (i.e.entire female batch fecundity range in Figure 6b) a negative relationship between ov and bf is strongly suggested.

Oocyte size and environmental conditions
the results of our study agree with the idea that environmental temperature could play an important role in oocyte number and size regulation in fish populations.in this study the smaller oocyte size and higher batch fecundity in females from the northern area coincided with higher mean sea surface temperatures (4°C) than in the southern area.Examples of such relationships have been provided for different Clupeiform species.de Ciechomski (1973), for instance, found a positive latitudinal gradient in the size of Engraulis anchoita eggs collected from the plankton along the west coast of the south atlantic; this gradient was largely attributed to the decreasing temperatures at higher latitudes.tanasichuk and Ware (1987), alternatively, reported positive correlations between Clupea harengus egg size and temperature anomalies observed from 90 to 60 days before spawning along the northwest coast of north america.tascheri and Claramunt (1996) suggested that temperature is the main regulating factor in the inter-seasonal variations in oocyte size observed in Sardinops sagax off northern Chile.along the Japanese coast, temperature was also reported to be positively correlated with fecundity and negatively correlated with egg size for Engraulis japonicus (Funamoto and aoki, 2002), indicating the existence of a trade-off between fecundity and egg size, and that temperature probably modifies the energy levels dedicated to reproduction.this phenomenon was also seen in a different type of fish species: three Pomacentrus coelestis populations distributed in reefs along Japan's Pacific coast showed increased egg numbers and decreased egg sizes with higher sea temperatures (Kokita, 2003).
along with potential temperature effects, food availability for adults and larvae may also play a role in determining differences in oocyte number and size. in this study the females in the northern area showed a higher mean weight than females from the southern area at a same fish length, suggesting that northern females had been exposed to better feeding or growth conditions.also, compared to northern Chile, the possibilities for first feeding larvae to find food are lower in the southern spawning area because of lower concentrations of microplankton and higher turbulence in the water column resulting from winter storms (Castro et al., 2002;llanosrivera and Castro, 2004).temperature is also lower and, hence, the duration of the larval stages is longer in the south (tarifeño et al., 2008).therefore, in this area, larger eggs from which larger larvae hatch with a larger amount of yolk should be favoured (llanosrivera and Castro, 2006).Consequently, the differences in egg size between the northern and southern Chilean anchoveta populations could result from local adaptations to the environmental conditions in each area and may be part of a reproductive strategy to produce an optimal number of eggs and egg sizes in order to maximize young offspring survival in each spawning habitat (Johnston and leggett, 2002;Kokita, 2003).Ferrada et al. (2002) used molecular genetic analyses to determine that E. ringens belongs to a pure panmictic stock along the Chilean coast, although the presence of some morphs in low frequency could represent variations at a genetic level.a recent work (Valdivia et al., 2007) used studies of the parasitic fauna of E. ringens to show marked differences in the parasitic loads in individuals off northern and southern Chile, suggesting that some degree of separation should exist, at least on an ecological level.thus, the cumulative information to date suggests that evidence of a potential genetic divergence able to explain differences in reproductive characteristics such as oocyte or egg size are not clear.Consequently, the differences observed seem to be a result of the high phenotypic plasticity of the anchoveta in different reproductive environments.in summary, our results from this study suggest that some reproductive parameters differ between Engraulis ringens females from northern and southern Chile, and also that differences previously observed in the free planktonic egg stage originated inside the females prior to oocyte hydration.the differences in reproductive traits agree with adaptive points of view that suggest these result from local adaptations to the dominant conditions in each habitat, favouring the survival of young offspring, the presence of this variability in the reproductive characteristics suggests that the anchoveta is capable of a marked phenotypic plasticity (hence, the characteristics are not fixed for each population or stock) and that, for now, it is not possible to conclude that these differences are the result of any degree of genetic isolation conducive to genetic differentiation.aCKnoWlEdGEMEnts this study was supported by the Chilean national Commission of research and technology (FondECYt Projects no.1030819 and 1070502 to lr Castro and G Claramunt). Histological preparations were facilitated by the laboratory of dr.Gustavo Herrera and Gabriel Claramunt, M.sc., of the universidad arturo Prat and correspond to samples collected within the framework of a project financed by the Fishing investigation Fund (FiP). the first author (El) also thanks iFoP for facilities and support in the final stage of this research and the Graduate school of the universidad de Concepción for co-financial support during his studies in the Master of sciences Programme in Fisheries of the department of oceanography.He would also like thank J.C. Quiroz for providing support in data analysis.the senior author (lC) was partially supported by FondaP-CoPas during the last part of this study.

Fig. 2 .
Fig. 2. -length frequency distribution of E. ringens females from the northern and southern Chile spawning areas used for oocyte volume and batch fecundity analyses.

Fig. 3 .
Fig. 3. -relationship between female total length and total weight (natural logarithm) E. ringens of northern (empty circles) and southern Chile (solid circles).

Fig. 4 .
Fig. 4. -size frequency distribution of E. ringens oocytes in northern and southern Chile spawning areas obtained during the peak month of spawning 2002, 2003 and 2004.

Fig. 5 .
Fig. 5. -Median E. ringens oocyte volume (horizontal lines inside the box) in females collected from northern (1) and southern Chile (2) spawning areas in the winters of 2002, 2003 and 2004.range observed value (vertical doted lines on box) and outliers (points on graphs) are also given.

Fig. 6 .
Fig. 6. -a) relationship between female ovary free weight and ovary weight (natural logarithm) from northern (empty circles) and southern Chile (solid circles) E. ringens.b) relationship between batch fecundity and oocyte volume for northern (empty circles) and southern (solid circles) Chile.

Table 1 .
-results of a t-test to assess whether differences in oocyte volume occurred between the northern and south Chilean E. ringens spawning areas during the winters2002, 2003 and 2004.

Table 2 .
-one way anoVa to test differences in oocyte volume among years within each spawning area.

Table 3 .
-result of a two-way anoVa to test effect of population and year on oocyte size E. ringens .